carbon dioxide removal from anaesthetic gas circuits using ... integr… · carbon dioxide removal...

Ana Filipa Fernandes Vaz Portugal

Carbon Dioxide Removal from Anaesthetic Carbon Dioxide Removal from Anaesthetic Carbon Dioxide Removal from Anaesthetic Carbon Dioxide Removal from Anaesthetic

Gas Circuits Using Absorbent Membrane Gas Circuits Using Absorbent Membrane Gas Circuits Using Absorbent Membrane Gas Circuits Using Absorbent Membrane

ContactorsContactorsContactorsContactors

Dissertation presented for the degree of

Doctor of Philosophy in Chemical and Biological Engineering

by

University of Porto

Supervisors:

Adélio Miguel Magalhães Mendes

Fernão Domingos de Montenegro Baptista Malheiro de Magalhães

LEPAE −−−− Chemical Engineering Department Faculty of Engineering

University of Porto

Porto, February 2009

Acknowledgements

I would like to express my gratitude to the Portuguese Foundation for Science and

Technology (FCT) for the PhD grant, reference SFRH/BD/16621/2004 and for the

financial support through the project POCTI/EQU/45182/2002. I am also grateful to the

European commission for the project Growth GRD1-2001-40257.

My acknowledgments go to my supervisors Prof. Adélio Mendes and Prof. Fernão de

Magalhães for giving me the opportunity and conditions to perform the present work,

for the scientific suggestions and recommendations and for the trust and

encouragement.

I am grateful to everybody from the OOIP group, in the Netherlands for the hospitality,

for the generosity on sharing their knowledge and for their skilled technical support.

Special thanks to Peter and Prof. Geert Versteeg - your teachings were essential for the

proceeding of this work. Thanks very much also to the true friends I made there which

never hesitate to aid me whenever I needed!

My gratitude goes also to my colleagues from the lab for their help and contribution to

the present work, for their support and, of course, for standing me singing or

complaining during my experimental work.

I would like to expand this acknowledgement to the staff of LEPAE (spetially

LEPAE/AMP) and of the Chemical Engineering department at FEUP for their kindness

and assistance.

Thanks very much to the people from E319 for the counsels and encouragement and

above all, for the good humour and friendship extended far beyond labour time... you

provided the best working environment I could ever find!

Thanks to all my friends; those which helped me with fruitful discussions and

recommendations, those with whom I’ve shared my PhD quotidian, and those which,

even when not directly related to this task, made me keep the necessary confidence and

good mood to move further. Special thanks to Tiago and his family who was also my

family during a significant part of this process.

For their unconditional support and for being so very special, I would like to

acknowledge my family. Particularly, I thank my aunts Maria Luisa Portugal Basílio, on

the subject “Carbon dioxide capture and sequestration”, Ana Portugal Crespo de

Carvalho, on the subject “Anaesthesia”, and Ana Paula Vaz Fernandes, on the subject

“How to deal with a PhD”.

Finally, always above all and above everything, thanks to my closest family: my Father,

my Mother and André, for supporting, for believing, for being this amazing example

that I am so proud of... for all possible reasons. Anything or any me would ever be

possible without you!!!

Preface

The present work was carried out at the Laboratory of Processes, Environmental and

Energy Engineering (LEPAE), in the Chemical Engineering Department of the Facu.lty

of Engineering – University of Porto (FEUP), between 2003 and 2009, under the

framework of the projects POCTI/EQU/45182/2002 (funded by the Fundação para a

Ciência e Tecnoologia) and European Growth Project GRD1-2001-40257 – SpecSep

(funded by the European Commission). This thesis contains different papers that were

written and published or submitted for publication in international journals during the

development of the PhD work.

I

Contents

Figure captions ……………………………………………………………. V Table captions …………………………………………………………….. XI Abstract …………………………………………………………………….. XV Sumário ………… ………………………………………………………….. XVII Résumé …………………………………………………………………… ... XX

Part I

1. Introduction ………………………………………………………. 3

1.1. Anaesthesia………………………………………………….. 3

1.2. Hollow Fiber absorbent Membrane Contactors………….. 6 1.3. Selection of Liquid Absorbents for CO2 Removal from

Anaesthetic Gas Circuits …………………………............. 10 1.4. Motivation and Outline of the Thesis……………………… 12 1.5. References………………………....................................... 15

Part II

2. Characterization of potassium glycinate for carb on

dioxide absorption purposes ……………………….............. 25

Abstract……………………………………………………............ 25

2.1. Introduction………………………………………................. 26

2.2. Zwitterion Reaction Mechanism ………………………...... 27

2.3. Mass Transfer................................................................... 30

2.4. Experimental…………………………………….................. 32

2.5. Results and Discussion ………………………………........ 36

2.6. Conclusions………………………………........................... 49

2.7. Nomenclature………………………………………............. 49

2.8. References………………………………………................. 52

2.A. Appendix - Experimental kinetic data…………………...... 56

II

3. Carbon dioxide absorption kinetics in potassium

threonate.......................................... ...................................... 65

Abstract……………………………………………...................... 65

3.1. Introduction……………………………………..................... 66

3.2. Reaction Mechanism........................................................ 67

3.3. Mass Transfer……………………………………................ 69

3.4. Physical Properties........................................................... 70

3.5. Experimental……………………………………………........ 72

3.6. Results and Discussion ……………………………………. 76

3.7. Conclusions……………........……………………............... 86

3.8. Nomenclature……….…………………………………......... 87

3.9. References………………………………………................. 89

3.A. Appendix - Experimental kinetic data............................... 94

Part III

4. Solubility of carbon dioxide in aqueous solution s of

amino acid salts ……………………………………………....... 101

Abstract……………………………………………………............ 101

4.1. Introduction………………………………………………...... 102

4.2. Modelling.......................................................................... 104

4.3. Experimental…………………………………………........... 108

4.4. Results and Discussion…………………………………...... 110

4.5. Conclusions………………………………………………..... 126

4.6. Nomenclature………………………………………………... 127

4.7. References………………………………………………....... 128

Part IV

5. Carbon dioxide removal from anaesthetic gas circ uits

using absorbent membrane contactors with amino acid

salt solutions ……………………………………...................... 137

Abstract…………………………………………………............... 137

5.1. Introduction……………………………………..................... 138

III

5.2. Mass Transfer with Chemical Reaction............................ 140

5.2.1. Chemical reaction........................................................... 140

5.2.2. Analogy to conventional mass transfer models.............. 142

5.2.3. Mathematical model........................................................ 146

5.2.4. Numerical resolution strategy......................................... 151

5.3. Results and Discussion.................................................... 152

5.3.1. Model validation.............................................................. 152

5.3.2. Performance of a membrane contactor for CO2 removal

from anaesthesia breathing circuits.......................................... 156

5.4. Conclusions.......…………………………………………...... 165

5.5. Nomenclature...........................................……………...... 166

5.6. References........…………………………………………...... 168

5.A. Appendix - Spatial discretization method......................... 175

Part V

6. General conclusions and Future Work ……………....... ....... 181

6.1. General Conclusions……………………………................. 181

6.2. Suggestions for Future Work…………………………........ 185

Appendix A. Details on the Experimental Setups Used ................... 187

V

Figure Captions

Figure 1.1 Schematic representation of the 2CO mass transfer in a hollow

fiber............................................................................................... 7

Figure 2.1 Simplified scheme of the experimental set-up............................. 34

Figure 2.2 2CON as a function of

2COP at 298 K for a potassium glycinate

concentration of 0.587 M............................................................. 36

Figure 2.3 Experimental Henry constants of 2N O in water and in

potassium glycinate solutions as a function of temperature.

Comparison with the solubility in water determined by

Versteeg and Van Swaaij (1988).................................................. 38

Figure 2.1 Parity plot of experimental enhancement factor and the

DeCoursey approximation........................................................... 45

Figure 2.5 Overall absorption kinetic constant as a function of potassium

glycinate concentration and for different temperatures:

experimental values and model lines. Solid lines correspond to

the model that takes into account the ionic strength and dashed

lines to the zwitterion model........................................................ 47

Figure 2.6 Apparent absorption kinetic constants as a function of

potassium glycinate concentration and at different

temperatures: experimental values and model lines. Solid lines

correspond to the model that takes into account the ionic

strength and dashed lines to the zwitterion model........................ 47

Figure 2.7 Figure 2.7 - Brønsted plot of Penny and Ritter (1983) at 293,

298 and 303 K – Comparison with the present work................... 48

Figure 3.1 Chemical structure of potassium threonate.................................. 67

Figure 3.2 Experimental set-up sketch.……………………….……………. 75

Figure 3.3 Sechenov plots of the 2N O solubility in potassium threonate

solutions........................................................................................ 78

VI

Figure 3.4 Threonate anion specific parameter as a function of

temperature................................................................................... 79

Figure 3.5 Comparison of 2CO absorption flux in potassium threonate,

potassium glycinate and diethanolamine (DEA) solutions at 1

M and 298 K (all measurements were performed in the setup

presented in Figure 3.2)................................................................ 82

Figure 3.6 Logarithmic plot of the overall absorption kinetic constant as a

function of the potassium threonate concentration -

Experimental values and model curves........................................ 85

Figure 3.7 Semi-log plot of the apparent absorption kinetic constant,

app ov Sk k C= , as a function of the solution ionic strength -

Experimental values and model curves........................................ 85

Figure 4.1 Experimental set-up sketch........................................................... 108

Figure 4.2 Semi-log plot of the solubility of 2CO in aqueous solutions of

MEA 2.5 M, at 313 K - comparison with results from

literature………………………………………………………… 110

Figure 4.3 Semi-log plot of the experimental solubility of 2CO in aqueous

solutions of potassium glycinate, 1.0 -3mol dm⋅ - comparison

with the results from Song et al. (2006) for an aqueous solution

of sodium glycinate 1.06 -3mol dm⋅ , at 313 and 323 K............... 111

Figure 4.4 Semi-log plot of the experimental solubility of 2CO in 3.0 M

aqueous solutions of potassium glycinate - comparison with the

results from Song et al. (2006) for an aqueous solution of

sodium glycinate 3.09 M, at 303, 313 and 323 K......................... 115

Figure 4.5 Solution loading as a function of the 2CO equilibrium partial

pressure in aqueous solutions of potassium glycinate at 313 K -

comparison with MEA at 2.5 M. Solid lines are provided to

make the figure clearer and do not correspond to theoretical

model results................................................................................. 117

VII

Figure 4.6 Semi-log plot of the experimental solubility of 2CO in aqueous

solutions of potassium threonate and potassium glycinate with

concentrations 1.0 M at 313 K..................................................... 118

Figure 4.7 Effect of changing the carbamate hydrolysis and amine

deprotonation equilibrium constants independently on the

predicted 2COP versus loading curves…………………………... 119

Figure 4.8 Effect of changing the carbamate hydrolysis and amine

deprotonation equilibrium constants simultaneously on the

predicted 2COP versus loading curves…………………………... 119

Figure 4.9 Solubility of 2CO in potassium glycinate solutions at 293 K –

experimental values and model curves…………………………. 124

Figure 4.10 Parity plot of the predicted and experimental loadings of 2CO

in solution for all data analysed………………………………… 124

Figure 4.11 Species concentrations as a function of loading for a potassium

glycinate solution, 1.0 M, at 313 K obtained using the

Deskmukh-Mather model. Note that points are not

experimental data but simulation results……………………….. 125

Figure 5.1 Sketch of a closed anaesthetic breathing circuit using hollow

fiber membrane contactors for 2CO removal............................... 139

Figure 5.2 Absorption flux of 2CO in water as a function of Gz –

numerical model (NM) and conventional model (CM) results.

Simulation conditions: 0.196ε = , 7/ 5.818 10L MGz Rτ⋅ = × ,

0.833Am = , -3, 40.34 mol mg TC = ⋅ .............................................. 152

Figure 5.3 E vs Ha plot for reactions (34), (35) and(36) – numerical

(NM) and conventional (CM) models results. Simulation

conditions: 412.63Gz= , / 19.795L GRτ = , 5/ 1.36 10L MRτ = × ,

0.196ε = , 1Am = , * 1BD = , -3, 41.6 mol mg TC = ⋅ , , 5MB feedC =

and 0.8K = .................................................................................. 154

VIII

Figure 5.4 Radial profiles at the fiber outlet for pseudo first order (PFO)

and instantaneous reaction (IR) regimes, for a direct second

order reaction – equation (34). Simulation conditions: Laminar

flow, 412.63Gz= , / 19.795L GRτ = , 5/ 1.36 10L MRτ = × ,

0.196ε = , 1Am = , * 1BD = , -3, 41.6 mol mg TC = ⋅ , , 5MB feedC = . 155

Figure 5.5 Radial profiles at the fiber outlet for pseudo first order (PFO)

and instantaneous reaction (IR) regimes, for a direct second

order reaction – equation (35). Simulation conditions: Laminar

flow, 412.63Gz= , / 19.795L GRτ = , 5/ 1.36 10L MRτ = × ,

0.196ε = , 1Am = , * 1BD = , -3, 41.6 mol mg TC = ⋅ , , 5MB feedC =

and 0.8K = .................................................................................. 155

Figure 5.6 Axial profiles along the contactor for co- and counter-current

operation and for different LQ and 2 ,RNH feedC . Simulation

conditions: 0.196ε = , Rshell= 2 × 10−2 m ..…………………….. 160

Figure 5.7 Influence of the contact area on the 2CO molar fraction at the

contactor exit for different 2 ,RNH feedC and LQ and for counter-

current operation........................................................................... 161

Figure 5.8 Influence of the amino acid salt feed concentration liquid flow

rate on the 2CO concentration at the contactor exit for different

LQ and for co- and counter-current operations –

20.8796 mA = . Lines are for improving the read........................ 163

Figure 5.9 Influence of the liquid flow rate on the 2CO concentration at

the contactor exit for different 2 ,RNH feedC and for co- and

counter-current operations – 20.8796 mA = . Lines are for

improving the read........................................................................ 164

Figure 5.A1 Schematic representation of the spatial discretization and cell

mass balance................................................................................. 176

IX

Figure A1 Setup used for the physical absorption and kinetics

measurements of 2CO in potassium glycinate (Chapter 2) - the

gas vessel and pressure controller are located behind the panel.. 188

Figure A2 Detail of the setup - stirred reactor (liquid volume: 600 cm3,

reactor diameter: 9.09 cm)............................................................ 189

Figure A3 Setup used for the determination of the physical absorption and

reaction kinetics of 2CO in potassium threonate (Chapter 3)

and for the equilibrium measurements of 2CO in potassium

glycinate (Chapter 4) - liquid volume: 50 cm3, reactor diameter:

3.87 cm......................................................................................... 190

Figure A4 Comparison of the experimental results obtained using the

setup at Porto University and using the setup from Twente

University..................................................................................... 191

XI

Table Captions

Table 1.1 Structural formulas of the amino acids characterized in the

present dissertation...................................................................... 14

Table 2.1 Densities of potassium glycinate solutions - ( )-3kg mρ ⋅ ……… 37

Table 2.2 Experimental Henry constants of 2N O in potassium glycinate

solutions....................................................................................... 37

Table 2.3 Sechenov’s constants for solubility of 2N O in aqueous

potassium glycinate solutions...................................................... 39

Table 2.4 Sechenov’s constants for solubility of 2CO in aqueous

potassium glycinate solutions...................................................... 40

Table 2.5 Henry constants of 2CO in potassium glycinate solutions

computed based on the Sechenov’s model -

( )2

3 -1Pa m molCOH ....................................................................... 40

Table 2.6 Viscosity and diffusivity of 2N O and 2CO in potassium

glycinate solutions........................................................................ 41

Table 2.7 Experimental values of the overall kinetic constant assuming

pseudo-first order behaviour........................................................ 42

Table 2.8 Computed values of SD used to calculate E∞ -

( )10 2 -110 m sSD × ⋅…………………………………………….... 44

Table 2.9 Ha and minimum values of E∞ used for computing ovk

assuming PFO.............................................................................. 44

Table 2.10 Experimental values of the overall kinetic constants of

potassium glycinate calculated using the DeCoursey equation -

( )-1sovk ......................................................................................... 45

Table 2.A1 Kinetic data of the reaction of 2CO with potassium glycinate at

0.0994 M and 298 K..................................................................... 57

XII


0.299 M and 293 K....................................................................... 57


0.299 M and 298 K....................................................................... 58


0.299 M and 303 K....................................................................... 58


0.587 M and 293 K....................................................................... 59


0.587 M and 298 K....................................................................... 59


0.587 M and 303 K....................................................................... 60


0.999 M and 293 K....................................................................... 60


0.999 M and 298 K....................................................................... 61


0.999 M and 303 K....................................................................... 61


1.984 M and 293 K....................................................................... 62


1.984 M and 298 K....................................................................... 63


1.984 M and 303 K....................................................................... 63


3.005 M and 298 K....................................................................... 64

Table 3.1 Densities and viscosities of potassium threonate solutions…...... 77

XIII

Table 3.2 Henry constants of 2N O and 2CO in water and in potassium

threonate solutions. All values are experimental except for 2CO

in potassium threonate solutions that were computed based on

Sechenov’s model - ( )3 1Pa m molH −⋅ ⋅ ………………………... 77

Table 3.3 Sechenov constants and specific parameters of Schumpe model

for the solubility of 2N O and 2CO in potassium threonate

solutions....................................................................................... 79

Table 3.4 Diffusion coefficient of 2N O , 2CO and potassium threonate in

potassium threonate solutions computed based on the Stokes-

Einstein relation - ( )10 2 110 m sD −× ⋅ ............................................ 80

Table 3.5 Physical mass transfer coefficient of 2CO in potassium

threonate solutions, computed based on equation (20) -

( )6 -110 m sLk × ⋅ ............................................................................ 81

Table 3.6 Experimental overall kinetic constants using the PFO and the

DeCoursey (DC) approaches - ( )1sovk − ....................................... 83

Table 3.A1 Flux of 2CO in 3 M potassium threonate solutions as a function

of the 2CO partial pressure, at 298 K........................................... 94

Table 3.A2 Flux of 2CO in 0.1 M potassium threonate solutions as a

function of the 2CO partial pressure, at 293, 298 and 303 K…... 95






of the 2CO partial pressure, at 293, 298 and 303 K……………. 96


of the 2CO partial pressure, at 293, 298, 303 and 313 K............. 97

XIV

Table 4.1 Solubility of 2CO in aqueous solutions of MEA 2.5 M............... 111

Table 4.2 Experimental solubility of 2CO in aqueous solutions of

potassium glycinate 0.1 M........................................................... 112






potassium threonate 1.0 M and 313 K.......................................... 117

Table 4.6 Equilibrium constants of reactions (1) to (5) and Henry

coefficient of 2CO in potassium glycinate solutions................... 120

Table 4.7 Effective size of the hydrated ions, based on the work by

Kielland (1937). .......................................................................... 121

Table 4.8 Model parameters fitted for the system potassium glycinate-

water- 2CO .................................................................................... 123

Table 5.1 Physical and chemical parameters used to model the absorption

of 2CO in potassium glycinate aqueous solutions in a hollow

fiber membrane contactor - liquid phase concentrations are

expressed in molarity................................................................... 158

Table 5.2 Simulation conditions................................................................... 159

XV

Abstract

The present dissertation concerns the study of hollow fiber absorbent membrane

contactors and their application for carbon dioxide removal from anaesthetic closed

breathing circuits.

Carbon dioxide removal from closed anaesthetic circuits is currently achieved using

canisters containing mixtures of alkali hydroxides. However, the volatile anaesthetics

react exothermally with these absorbents, generating potentially harmful products such

as carbon monoxide and compound A; besides, the exhausted canisters are

contaminated hospital waste, dangerous and expensive to treat. This work proposes to

contribute for the development of a safer and more environmentally friendly technology

for carbon dioxide removal, based on membrane contactors using regenerable absorbent

solutions. These solutions should be able to absorb carbon dioxide fast and reversibly

when it is present in low concentrations (carbon dioxide concentration must be reduced

from 5 to 0.5 % in an anaesthesia loop).

Although hollow fiber absorbent membrane contactors have been widely used and

studied for carbon dioxide absorption purposes, there are two main difficulties that must

be overcome to make them suitable for the suggested application. Firstly, dense

membranes are necessary (instead of the porous membranes generally used in these

devices) to avoid the transmission of pathogenic microorganisms from the breathing

circuit to the absorbent solution. Additionally, new absorbent solutions need to be

developed, since the ones commonly used (aqueous solutions of alkanolamines)

undergo oxidative degradation in highly oxygenated environments, originating toxic

compounds. The present dissertation is predominantly focussed on the latter problem.

Aqueous solutions of amino acid salts overcome some of the drawbacks associated to

the use of alkanolamines. Among the available alkali salts of amino acids, potassium

glycinate was chosen as a model for the subsequent studies, since glycine is the simplest

amino acid and it has a relatively low cost. Additionally, the molecular structure of

potassium glycinate indicates that high absorption kinetics towards carbon dioxide

might be expected. Potassium salt of threonine was also studied for comparison and

because its molecular structure envisioned better regeneration properties.

XVI

In order to estimate the diffusion coefficients of both carbon dioxide and amino acid

salt, densities and viscosities of potassium glycinate and potassium threonate aqueous

solutions were experimentally measured. The physical solubility of carbon dioxide in

aqueous solution was determined based on the nitrous oxide analogy. Therefore, the

solubility of this gas in aqueous solutions was experimentally obtained.

The kinetics of the reactions of carbon dioxide with potassium glycinate and potassium

threonate were determined using a stirred reactor with a flat gas-liquid interface. The

results were interpreted using the DeCoursey approach and an expression for the rate of

absorption as a function of temperature and solution concentration was derived for each

amino acid salt, based on the zwitterion reaction mechanism. It was observed that

potassium glycinate absorbs carbon dioxide faster than potassium threonate and, for

both amino acid salts, the absorption rate is strongly dependent on the solution ionic

strength.

Solubility of carbon dioxide in potassium glycinate aqueous solutions was determined

in a stirred reactor at different temperatures, amino acid salt concentrations and carbon

dioxide partial pressures. Absorption equilibrium data was further interpreted using the

Deshmukh-Mather and the Kent-Eisenberg models. For potassium threonate, the carbon

dioxide solubility was also measured, but for a limited set of conditions. This amino

acid salt showed lower absorption capacity than potassium glycinate.

A bi-dimensional model was developed to evaluate the carbon dioxide removal

performance of a hollow fiber membrane contactor. The model considered potassium

glycinate solutions as absorbents and a composite membrane, made of a porous support

layer and a dense thin layer. Both co- and counter-current operations were analysed.

The influence of some system parameters on the separation achieved was studied.

The use of hollow fiber absorbent membrane contactors with amino acid salt solutions

was found to be suitable for carbon dioxide removal from closed anaesthetic breathing

circuits.

XVII

Sumário

A presente dissertação versa sobre o estudo de contactores absorvedores de membranas

de fibras ocas e sua aplicação na remoção de dióxido de carbono de circuitos

anestésicos fechados.

A remoção de dióxido de carbono de circuitos anestésicos fechados é correntemente

levada a cabo usando recipientes contendo misturas de hidróxidos alcalinos. No entanto,

os compostos voláteis anestésicos reagem exotermicamente com estes absorventes

formando produtos potencialmente perigosos como o monóxido de carbono e o

composto A. Adicionalmente, depois de saturados, os recipientes são lixo sólido

hospitalar contaminado que requer tratamentos caros e perigosos. Pretende-se que este

trabalho contribua para o desenvolvimento de uma tecnologia mais segura e amiga do

ambiente para a remoção de dióxido de carbono, baseada no uso de contactores de

membrana com soluções absorventes recicláveis. Estas soluções devem conseguir

remover o dióxido de carbono rápida e reversivelmente, quando este se encontra pouco

concentrado na corrente gasosa (a concentração de dióxido de carbono deve ser

reduzida de 5 para 0.5 % em cada ciclo anestésico).

Apesar dos contactores de membranas de fibras ocas terem sido vastamente estudados e

utilizados com o propósito de absorver dióxido de carbono, duas limitações essenciais

têm que ser ultrapassadas para que estes se tornem adequados para a aplicação sugerida.

Em primeiro lugar, são necessárias membranas densas (em vez das membranas porosas

habitualmente utilizadas nestas unidades) de modo a evitar a transmissão de micro

organismos patogénicos do circuito respiratório para a solução absorvente.

Adicionalmente, é necessário que novas soluções absorventes sejam desenvolvidas, uma

vez que, as geralmente usadas (soluções aquosas de alcanolaminas) oxidam em

ambientes muito oxigenados, originando compostos tóxicos. A presente dissertação

foca-se predominantemente neste último problema.

As soluções aquosas de sais de aminoácidos superam algumas das limitações associadas

ao uso de alcanolaminas. De entre os sais alcalinos de aminoácidos, o glicinato de

potássio foi escolhido como modelo para um estudo mais aprofundado porque a glicina

é o aminoácido mais simples e tem um custo relativamente baixo. Adicionalmente, a

XVIII

estrutura molecular do glicinato de potássio indica que elevadas cinéticas de absorção

do dióxido de carbono são espectáveis. O treonato de potássio foi também estudado

para comparação e porque a sua estrutura molecular fazia prever maior

regenerabilidade.

Com o propósito de estimar os coeficientes de difusão do dióxido de carbono e do sal de

aminoácido, foram medidas experimentalmente a densidade e viscosidade de soluções

aquosas de glicinato de potássio e treonato de potássio. A solubilidade física do dióxido

de carbono nas soluções aquosas foi determinada através da analogia com o protóxido

de azoto. Para tal, a solubilidade deste gás nas soluções aquosas foi obtida

experimentalmente.

As cinéticas da reacção do dióxido de carbono com o glicinato de potássio e o treonato

de potássio foram determinadas num reactor agitado com uma interface gás-líquido

plana. Interpretaram-se os resultados através da aproximação de DeCoursey e, para cada

sal de aminoácido, foi derivada uma expressão relacionando a velocidade de absorção

com a temperatura e a concentração da solução, baseada no mecanismo de zwitterion.

Observou-se que o glicinato de potássio absorve dióxido de carbono mais depressa do

que o treonato de potássio e que, para ambos os sais, a velocidade de absorção é

fortemente dependente da força iónica da solução.

A solubilidade do dióxido de carbono em soluções aquosas de glicinato de potássio foi

medida num reactor agitado para diferentes temperaturas, concentrações de sal de

aminoácido e pressões parciais de dióxido de carbono. Os resultados do equilíbrio de

absorção foram seguidamente interpretados usando os modelos de Deshmukh-Mather e

de Kent-Eisenberg. A solubilidade do dióxido de carbono em treonato de potássio foi

também medida, mas para um conjunto de condições limitado. Este sal de aminoácido

apresentou menor capacidade de absorção do dióxido de carbono que o glicinato de

potássio.

Foi desenvolvido um modelo bidimensional para avaliar o desempenho na remoção de

dióxido de carbono de um contactor de membranas de fibras ocas. No modelo,

consideraram-se soluções de glicinato de potássio como absorventes e uma membrana

compósita, constituída por uma camada de suporte poroso e uma fina camada densa.

XIX

Operações em co- e contra corrente foram analisadas. Estudou-se a influência de alguns

parâmetros do sistema na separação conseguida.

Concluiu-se que os contactores absorvedores de membranas de fibras ocas são uma

tecnologia viável para remover dióxido de carbono de circuitos anestésicos fechados.

XX

Résumé

Ce travail se penche sur l’étude des contacteurs absorbeurs de membranes de fibres

creuses et son application dans l’extraction du dioxyde de carbone dans les circuits

fermés d’anesthésie.

L’extraction du dioxyde de carbone dans les circuits anesthésiques fermés est mise en

œuvre à l’aide de récipients contenant des mélanges d’hydroxydes alcalins. Cependant,

les composés volatiles anesthésiques ont une réaction exothermique avec ces absorbants

provocant la formation des produits dangereux comme le monoxyde de carbone et le

composant A. De plus, après saturation, les récipients utilisés se transforment en déchet

toxique, qui demande un traitement cher et dangereux. L’objectif de cette étude est de

contribuer au développent d’une technologie plus sécurisée et respectueuse de

l’environnement pour l’extraction du dioxyde de carbone, basé sur l’utilisation des

contacteurs de membrane avec des préparations d’absorbants recyclables. Ces

préparations devront permettre d’extraire le dioxyde de carbone d’une façon rapide et

réversible, lorsque le dioxyde de carbone est présent dans le gaz à faible concentrations

(la concentration de dioxyde de carbone doit être réduite de 5 à 0.5% par cycle

anesthésique).

Bien que les contacteurs de membranes de fibres vides aient déjà largement été étudiés

et utilisés pour l’extraction du dioxyde de carbone, il reste deux principales contraintes à

résoudre pour cette application. Premièrement, il est nécessaire d’utiliser des

membranes denses (au lieu des membranes poreuses généralement utilisées dans ces

outils) afin d’éviter la transmission de micro-organismes pathogènes du circuit de

respiration aux préparations absorbantes. Deuxièmement, il reste à développer de

nouvelles préparations absorbantes, vu que celles actuellement utilisées (solutions

aqueuses d’alkanolamines) subissent une dégradation oxydative en milieu

particulièrement oxygénés, provoquant l’apparition de composés toxiques. Cette étude

se concentre principalement sur cette deuxième contrainte.

Des solutions de sels d’acides aminés permettent de dépasser les limites associées à

l’utilisation des alkanolamines. Parmi les sels alcalins d’acides aminés disponibles, le

XXI

glycinate de potassium a été choisie comme modèle pour cette étude, car la glycine est

l’acide aminé le plus simple et a un prix raisonnable. De plus, la structure du glycinate

de potassium montre que l’on peut s’attendre à des cinétiques d’absorption du dioxyde

de carbone élevées. Nous avons étudié également le sel de potassium de thréonine, pour

comparaison d’une part, et du fait que sa structure moléculaire présentait de meilleurs

propriétés de régénération.

Afin d’estimer les coefficients de diffusion aussi bien du dioxyde de carbone que des

sels d’acides aminés, nous avons mesuré expérimentalement les densités et les

viscosités du glycinate de potassium et du thréonate de potassium. Nous avons

déterminé la solubilité physique du dioxyde de carbone en solution aqueuse par

analogie avec le protoxyde d’azote. Ce faisant, la solubilité de ce gaz en solution

aqueuse a été obtenu expérimentalement.

Nous avons déterminé les cinétiques de réaction du dioxyde de carbone avec le

glycinate de potassium et le thréonate de potassium grâce à un réacteur à agitation avec

une interface gaz-liquide. Les résultats ont été interprétés par une approche DeCoursey

et, pour chacun des acides aminés, nous avons dérivé une équation mettant en relation la

vitesse d’absorption en fonction de la température et de la concentration de la solution,

en nous basant sur un mécanisme de réaction zwiterrienne. Nous avons observé que le

glycinate de potassium absorbe le dioxyde de carbone plus rapidement que le thréonate

de potassium, et que, pour les deux sels d’acides aminés, le taux d’absorption est

particulièrement dépendant de la force ionique de la solution.

La solubilité du dioxyde de carbone dans une solution aqueuse de glycinate de

potassium a été déterminée dan un réacteur à agitation à températures, concentrations

d’acides aminés et pressions partielles de dioxyde de carbone différentes. Les données

d’équilibre d’absorption ont été interprétées par les modèles de Deshmukh-Mather et de

Kent-Eisenberg. Pour le thréonate de potassium, la solubilité du dioxyde de carbone a

également été mesurée, mais avec des combinaisons de conditions limitées. Cet acide

aminé a montré des capacités d’absorption plus faibles que le glycinate de potassium.

Un modèle bidimensionnel a été développé afin d’évaluer les performances d’extraction

du dioxyde de carbone par un contacteur de membrane à fibres creuses. Ce modèle

XXII

prend en compte le glycinate de potassium en tant qu’absorbant et une membrane

composite, obtenue par superposition de couches support poreuses et une fine couche

dense. Nous avons analysé aussi bien les co- et contre-courants, ainsi que l’influence de

certains systèmes de paramètres sur la séparation obtenue.

En conclusion, l’utilisation de contacteurs de membrane à fibres creuses avec des

solutions de sels d’acides aminés s’est révélé être adaptée à l’extraction du dioxyde de

carbone dans les circuits fermés de respiration d’anesthésie.

Part I

3

1. Introduction

The present dissertation aims to study a novel technology for carbon dioxide ( 2CO )

removal from anaesthetic closed breathing circuits. The viability of using hollow fiber

absorbent membrane contactors for this application is analysed and new absorbent

liquids are studied.

1.1. Anaesthesia

General anaesthesia is a technique to bring and keep the patient unconscious by the

administrations of drugs which can be provided intravenously or by inhalation (Brandi,

2008; Pontes, 2006). General anaesthesia provides analgesia (absence of pain), amnesia

(no memory), and muscle relaxation (Brandi, 2008). Apart from the heart, all the patient

body muscles become relaxed and therefore breathing must be externally induced.

During general anaesthesia, the patient is continuously ventilated with a gaseous

mixture typically composed of, approximately, 70 % carrier gas - usually nitrous oxide

( 2N O ) or air, 30 % oxygen (2O ) and 1 to 8 % volatile anaesthetic (Mendes, 2000;

Pontes, 2006). The volatile anaesthetics currently used are halogenated compounds such

as halothane, enflurane, isoflurane, desfluorane and sevofluorane being the latter two

the most widely used nowadays (Pontes, 2006; Whalen et al., 2005).

The anaesthetic gas mixture can be delivered to the patient in open or closed breathing

circuits and most of the anaesthetic machines can work using both circuits and switch

between them. In the open breathing circuit, fresh gas is transferred to the patient and

there is no recycling of the expired gases. This system results in high fresh gas flows -

minimum of 5 -1L min⋅ (Dosch, 2004) – and it is commonly used for initializing the

anaesthesia. Closed breathing circuit (also called low flow anaesthesia) consists in

leading the expired air, containing the unused anaesthetic gases, back to the patient in

the subsequent inhalation (Baum and Woehlck, 2003). Since the halogenated volatile

anaesthetics are expensive substances, their waste must be kept as low as possible. In

addition, 2N O and the halogenated compounds are potentially green house gases and

their release into the atmosphere should be minimized (Dingley et al., 1999; Pontes,

4

2006). These reasons make the closed breathing circuit the desirable anaesthetic

arrangement. However, the gaseous current coming out from the patient contains an

excess of 2CO (around 5 %) resulting from the patient breathing and an excess of 2N

(around 3 %) that was dissolved in the body tissues and is released during anaesthesia

due to the lower 2N concentration in the inhaled gas, that need to be removed and

replaced by fresh anaesthetic gas before carry the mixture back again to the patient

(Mendes, 2000). In routine clinical practice, the excess of 2N is eliminated by

periodically venting the system (Pontes, 2006; Reinelt et al., 2001) and the 2CO

removal is currently achieved using soda lime (a mixture of calcium, potassium and

sodium hydroxides and water) or baralymeTM (a mixture of calcium, potassium and

octahydrated barium hydroxides) (Baum and Woehlck, 2003).

When Franz Kuhn first described a closed breathing circuit, in 1906, he raised concerns

about the potential harmful products resulting from the reaction of the volatile

anaesthetics with the 2CO absorbent (Baum and Woehlck, 2003). Compounds used for

anaesthesia substantially changed since then, however these concerns are still a reality

nowadays (Baum and Woehlck, 2003; Fan et al., 2008; Knolle and Gilly, 2000; Whalen

et al., 2005). Actually, all volatile anaesthetics react with conventional 2CO absorbents

when these become accidentally desiccated (Baum and Woehlck, 2003; Fan et al., 2008;

Knolle and Gilly, 2000; Whalen et al., 2005). Even when 2CO absorbents are properly

hydrated undesirable products are eventually formed during long time surgeries (longer

than four hours), although much less severe consequences are expected under this

conditions (Baum and Woehlck, 2003; Fan et al., 2008).

The volatile anaesthetic sevofluorane reacts with soda lime, especially during low flow

anaesthesia, generating, among others, the so called compound A - fluoromethyl-2,2-

difluoro-1-(trifluoromethyl)vinyl ether) - a potentially nephrotoxic compound (Baum

and Woehlck, 2003; Whalen et al., 2005). Carbon monoxide (CO) can be also

generated, especially when desfluorane is used (Baum and Woehlck, 2003; Fan et al.,

2008; Whalen et al., 2005). Besides the highly toxics compound A and CO, a number

of other degradation products can be formed upon the contact of the volatile anaesthetic

agents and the desiccated 2CO absorbents. These include methanol and formaldehyde

5

and other flammable gases not yet identified (Baum and Woehlck, 2003; Marini et al.,

2007). Considering the extreme heat produced in these reactions, there is a possibility of

ignition of these gases during the anaesthesia (Baum and Woehlck, 2003). To overcome

these drawbacks, a number of less reactive absorbents have been developed and some –

including DragerSorb FreeTM, AmsorbTM, LoFloSorbTM and SuperiaTM - present good

results concerning the generation of harmful reaction products even under desiccation

conditions (Baum and Woehlck, 2003; Marini et al., 2007; Murray et al., 1999; Struys et

al., 2004). However, these less reactive absorbents enable lower utilization times than

common soda lime, under comparable clinical conditions (Baum and Woehlck, 2003).

Besides the potentially harmful health effects, the environmental impact and the costs of

using the present technology for 2CO removal from anaesthetic gas circuits must be

taken into account. The exhausted 2CO absorbents are hospital solid waste, dangerous

and expensive to treat (Mendes, 2000). Usually, one canister of 1.5 L of absorbent is

enough to absorb 2CO over one week. However, the exhaustion of the absorbent may

occur before this period (Baum and Woehlck, 2003). For this reason, it would be

desirable to replace common 2CO absorbents by a safer but also cleaner technology.

Chemical absorption in liquid solutions is a proven and established technology to

perform 2CO separation from a variety of gas mixtures present in chemical process

industry (Idem and Tontiwachwuthikul, 2006; Ma'mun et al., 2007). A wide number of

studies related to the use of hollow fiber membrane contactors for 2CO separation have

been performed in the past few years (Al-Marzouqi et al., 2008; Kumar et al., 2002a; Li

and Chen, 2005; Rangwala, 1996; Yan et al., 2007). The process showed promising

results and is being implemented by several companies (Kumar, 2002).

In the present work the use of a hollow fiber absorbent membrane contactor for

continuous 2CO removal from anaesthetic gas circuits is proposed and its feasibility

studied. The membrane material used is considered dense (non-porous) and highly

permeable to 2CO , isolating the absorbent solution from the anaesthetic closed circuit.

After going through the 2CO absorption contactor, the absorbent solution is regenerated

in another contactor and sent back to the absorption contactor.

6

1.2. Hollow Fiber Absorbent Membrane Contactors

A membrane contactor is a device to bring in contact two different phases, for mass

transfer purposes, without dispersion of one phase into the other (Gabelman and

Hwang, 1999).

Porous membrane modules for gas absorption have been explored and successfully used

since 1975, when the technology was first proposed for blood oxygenators (Kumar,

2002) - nowadays, 99 % of the blood oxygenators sold in U. S. contain porous

membranes (Wickramasinghe et al., 2005). Many other application can be find such as

aeration of water in river and wastewater treatment plants, biological waste gas

treatment, removal of volatile organic compounds from water, aeration of shear-

sensitive cell cultures, aeration of reactors at high oxygen demand, removal of dissolved

oxygen in ultrapure water production, gas exchange in artificial gills, 2CO removal

from industrial gaseous streams, etc. (Vladisavljevic, 1999).

Usually, in a gas-liquid hollow fiber membrane contactor, the liquid flows inside the

fibers lumen and the gas flows in the shell (Li and Chen, 2005). The driving force for

the mass transfer is the concentration gradient between gas and liquid phases (Kumar,

2002; Li and Chen, 2005) and the process of mass transfer includes the following steps:

diffusion from the bulk of the gas to the outer membrane surface, diffusion trough the

membrane, dissolution in the liquid and diffusion accompanied (or not) by chemical

reaction in the liquid (Li and Chen, 2005).The selectivity is commonly provided by the

liquid and the membrane works as an interface between two media, although is possible

to use selective membranes (Li and Chen, 2005). A schematic representation of a

hollow fiber in a gas liquid membrane contactor for 2CO removal is shown in Figure

1.1.

7

Figure 1.1 – Schematic representation of the 2CO mass transfer in a hollow fiber.

Membrane contactors, and particularly hollow fiber membrane contactors, offer a

number of advantages when compared to traditional gas/liquid contactor devices such as

packed tower, spray tower, bubble column or venturi scrubber (among others)

(Gabelman and Hwang, 1999; Kumar, 2002; Li and Chen, 2005; Rangwala, 1996):

- Much larger contact area per unit volume - hollow fiber membrane contactors can

provide interfacial areas per unit volume around thirty times higher than other types of

contactors. Besides, since the two fluids flow independently, this surface area does not

depend on operational conditions such as the fluids flow rates.

- Because of the absence of interpenetration of the gaseous and liquid phases, these

apparatus do not present operational limitations like flooding, channeling, entrainment,

loading, weeping and foaming.

- Membrane modules can be linearly scaled up and, due to its modularity, different

separation capacities can be achieved by simply changing the number of modules used

and the contactor orientation is also not a matter of concern.

- Since membrane modules are compact, they are less energy consuming, and need

lower volumes to achieve identical separations being very interesting in an economical

point of view. They are also light on weight which makes them easy to be transported

and used for offshore applications.

- Aseptic operation is much easier to achieve than with other types of contactors, which

enables the process to be suitable for medical applications.

Hollow fiber membrane contactors have also some disadvantages (Gabelman and

Hwang, 1999; Kumar, 2002; Li and Chen, 2005; Rangwala, 1996):

8

- Due to the small diameter of fibers, the liquid flow inside the fibers is usually laminar.

As a consequence, the mass transfer coefficient is lower than in other types of

contactors. The membrane itself also provides an additional resistance to the mass

transfer.

- If the membranes to be used are porous, it must be assured that the pores are gas filled

during the mass transfer process. If the membrane is wetted, the mass transfer is greatly

penalized due to the presence of a stagnant liquid film in the membrane pores.

- Membranes are subject to fouling and they have a finite life time, which makes

necessary to change the modules from time to time.

- There is pressure drop along the module.

The use of hollow fiber membrane contactors for 2CO removal was first proposed by

Zhang and Cussler (1985a, 1985b) (Li and Chen, 2005) and since then a lot of research

on this particular application have been performed. This tremendous investment is due

to the role of separating the 2CO from flue gas for further sequestration (Idem and

Tontiwachwuthikul, 2006; Metz et al., 2005). The climate change due to the greenhouse

gases concentrations in the atmosphere is probably the most concerning environmental

problem at the present. 2CO is the greenhouse gas released in larger extend by

anthropogenic action (Idem and Tontiwachwuthikul, 2006; UNFCCC, 2008) and its

concentration in the atmosphere has risen by more than 30 % in the last 250 years

(Hampe and Rudkevich, 2003). Most of the 2CO emissions result from burning fossil

fuels (mainly coal and natural gas) to produce energy (Hampe and Rudkevich, 2003;

Metz et al., 2005) and the demand for energy is kept increasing in such a rate that

replacing the use of fossil fuels by renewable and clean energy sources would take more

time than we probably have to face this problem (CAETS, 1995). For this reason, there

is a huge interest on the development of technologies to capture and storage 2CO and

absorption on reactive solutions is pointed out as one of the most promising ones

(Favre, 2007). Relating to 2CO and oxygen, the composition of the exhausted flue gas

is very similar to that of the anaesthetic gas mixtures exhausted by the patient - namely,

the molar fraction of 2CO in the flue gas varies from 3 to 15 % and, in the exhausted

anaesthetic mixture, it is around 5 %. For this reason, it should be kept in mind most of

the research and analysis on the 2CO removal from flue gas using absorbent hollow

9

fiber membrane contactors can be applied to the 2CO removal from anaesthetic gas

circuits. The opposite is also true: the progresses achieved on the 2CO removal from

anaesthetic gas circuits using hollow fiber membrane contactors will eventually find

application on the flue gas treatment.

Generally, to model and to analyse the performance of the mass transfer between a gas

and an absorbent liquid in a hollow fiber membrane contactor, the following

information is required:

- The mass transfer coefficients of each component in the gas phase, which depend on

the flow hydrodynamics - unlike the fiber lumen, the hydrodynamics in the shell side is

usually difficult to describe. Several models have been proposed to describe gas flow in

the shell side (Keshavarz et al., 2008), but usually gas phase mass transfer coefficients

are experimentally obtained for the specific membrane module.

- The mass transfer coefficients trough the membrane, which depends on the membrane

characteristics (material, porosity, etc) and is membrane specific - usually, the

membranes used have high permeabilities and the resistance to mass transfer introduced

by the membrane is negligible.

- The physical solubility of the components in the liquid.

- The diffusion coefficients of the gas components and the reactive species in the

absorbent liquid – as mentioned before, the liquid flow inside the fiber lumen is usually

laminar and bi-dimensional models for diffusion or diffusion/reaction are necessary to

describe the mass transfer inside the liquid. Nevertheless, in some cases, especially for

physical absorption, it is possible to estimate a mass transfer coefficient in the liquid;

however the correlations used in these situations also require the previous knowledge of

the diffusion coefficients and liquid physical properties such as density and viscosity.

- Information about the reaction kinetics and equilibrium between the absorbent reactive

species and the reactive gases.

10

1.3. Selection of Liquid Absorbents for CO 2 Removal from

Anaesthetic Gas Circuits

An absorbent solution to be used in a membrane contactor for 2CO removal from

anaesthetic circuits must verify the following requirements: biocompatibility, low

vapour pressure (in order not to enter into the anaesthetic circuit), chemical and thermal

stability and compatibility with the membrane contactor, i.e. do not originate the

membrane swelling and, for porous membranes, do not wet the membrane pores.

Besides, it should present fast absorption and desorption kinetics and high absorption

capacity and it should be easily regenerable.

The absorbent solutions most widely used nowadays to separate 2CO from gaseous

mixtures in chemical industry are aqueous solutions of akanolamines and blends of

alkanolamines (Idem and Tontiwachwuthikul, 2006). Alkanolamines have been

extensively studied for 2CO absorption purposes and their aqueous solutions

characterized in detail concerning physical properties and reaction equilibrium and

kinetics towards 2CO (Austgen et al., 1989; Blauwhoff et al., 1984; Rochelle et al.,

2001; Versteeg et al., 1996; Versteeg and Van Swaaij, 1988; Weiland et al., 1993).

However, alkanolamines easily undergo oxidative degradation resulting in highly toxic

degradation products and this degradation is far more extensive in oxygen rich

environments (Goff and Rochelle, 2006; Holst et al., 2006; Kumar, 2002; Supap et al.,

2006). Besides, alkanolamines are organic substances with surface tensions

considerably lower than water and therefore they wet some commercially available

membranes (Kumar, 2002). For these reasons, alkanolamines might not be suitable for

2CO removal from anesthetic gas circuits and new absorbents, able to overcome these

drawbacks need to be developed.

Amino acids (or, more precisely, alkali salts of amino acids1) are being studied as a

possible alternative for alkanolamines (Feron and Jansen, 2002; Kumar et al., 2002b).

Amino acids have the same reactive group towards 2CO as alkanolamines and therefore

1 Amino acids exist in solution as a zwitterion (with the amine group protonated) -

1 2 3 1 2 3HOOC R R R N OOC R R R NH− +− − ↔ − − . It is necessary to make it react with an alkali hydroxide

(potassium hydroxide, for example) to enable it to react with 2CO .

11

they present equivalent equilibrium capacities and reaction kinetics (Holst et al., 2006;

Kumar et al., 2003a; Kumar et al., 2003b). However, due to their ionic nature, amino

acids present a number of advantages when compared to alkanolamines: they are much

more resistant to oxidative degradation and more thermally stable, present lower

volatilities (amino acids can be considered non volatile, so there is no loss of the active

specie during the process and no transfer to the anaesthetic circuit) and their solutions

have higher surface tensions (not wetting common and non expensive membranes) and

have densities and viscosities similar to water (which means that no extra hydrodynamic

concerns are introduced) (Feron and Jansen, 2002; Kumar et al., 2001; Kumar et al.,

2002b). Nevertheless, amino acids present a couple of drawbacks: they are more

expensive than alkanolamines and precipitation of the reaction products was observed

during the absorption of 2CO in their solutions (Hook, 1997; Kumar et al., 2003c;

Majchrowicz et al., 2006). Precipitation is a severe limitation if porous membrane

contactors are to be used because of possible blockage of membrane pores; even when

non porous membranes are used, hydrodynamic problems can arise because of

precipitation.

The general ability of an amino acid (or other amine based compound) to absorb 2CO is

determined by the molecular structure of the compound. There is a considerable amount

of information in literature relating the molecular structure of 2CO absorbents and

reaction characteristics such as absorption kinetics, equilibrium capacity and

regeneration extent (Caplow, 1968; Hook, 1997; Penny and Ritter, 1983; Sartori and

Savage, 1983; Singh et al., 2007).

2CO reacts with aqueous solutions of primary or secondary amines forming carbamates,

bicarbonates and carbonates (Caplow, 1968; Hook, 1997; Jensen et al., 1952).

Generically, the stability of the carbamates formed influence the absorption as follows:

amines which form stable carbamates react faster but present lower equilibrium

capacities at loadings higher than 0.5 2

-1CO AmAmol mol⋅ (i.e. for the same loading, the

2CO equilibrium pressure above the liquid is higher in solutions of this amines) and

they are more difficult to regenerate than amines which form unstable carbamates

(Hook, 1997; Park et al., 2003; Sartori and Savage, 1983). An amine is sterically

12

hindered when the amine group is connected to a secondary or terciary carbon, i.e. when

the carbon adjacent to the amine group is substituted. Sterical hindrance is known to

considerably reduce the carbamate stability (Sartori and Savage, 1983). Therefore,

sterically hindered amines present generally higher absorption capacities at high

loadings and show deeper desorption ability, but they exhibit lower absorption kinetics

when compared to their non-sterically hindered equivalents. In the same way, secondary

amines form less stable carbamates than primary amines. Singh et al. (2007) studied the

influence of the chain length on the amines’ absorption ability and concluded that

increasing the chain length does not bring any advantage to the absorption equilibrium

or to the absorption kinetics. Hook (1997) compared the regeneration achieved with

amines containing a potassium carboxilate group ( )2CO K− + (potassium amino acid

salts) and a hydroxymethylen group ( )2HOCH (amino alcohols) and concluded that the

amino alcohols enable higher desorption levels and at higher rates than the

corresponding amino acid salts.

Tertiary amines do not react directly with 2CO to form carbamates. Instead, they act as

a catalyst for the hydration of 2CO to form bicarbonate (Bonenfant et al., 2003; Sartori

and Savage, 1983). They are essentially slower absorbents than primary and secondary

amines but enable high absorption capacities (Bonenfant et al., 2003). Tertiary

alkanolamines are also easier to regenerate (Bonenfant et al., 2003; Derks et al., 2006).

They are often used blended with other amines which act as rate promoters (Bishnoi and

Rochelle, 2000; Derks et al., 2006).

1.4. Motivation and Outline of the Thesis

The doctoral work presented here arose in the context of the European Project entitled

“Development of New Materials and Processes to Enhance Specialty Gas Separations” -

SpecSep. The project concerned gas separations for medical applications and four

partners were involved on the subject of 2CO removal from anaesthetic gas streams and

from life support applications using hollow fiber absorbent membrane contactors:

LEPAE (Porto, Portugal), GKSS (Geesthacht, Germany), CSIC (Madrid, Spain) and

Drager AG (Luebeck, Germany).

13

In the framework of the SpecSep European Project, LEPAE was in charge of

characterizing different absorbents, focusing on the reaction kinetics and equilibrium

towards 2CO . Both commercially available amino acids and amino acids synthesized by

CSIC - Consejo Superior de Investigaciones Cientificas, Institute of Science and

Technology of Polymers - were considered. Since laboratory synthesis of absorbents is

an expensive and time consuming process and generally only a few grams can be

produced in each batch, a lot of effort was put on the development of a fast and

inexpensive methodology to analyze the performance of the absorbents using the lowest

amount of substance possible. For the pre-screening of the absorbents, a setup similar to

the ones used for the volumetric measurement of adsorption isotherms was used – this

setup and methodology are described in detail by Santos et al. (2008). In addition, the

solution was continuously stirred using a magnetic stirrer and the pressure decrease

inside the absorbent tank was recorded during the entire absorption process. Uptake

curves were drawn with the pressure decrease versus time. Using this method, only 10

mL of solution is spent for each experiment. Experiments were performed at 293 K.

During the SpecSep Project, the following commercially available amino acid salts

were pre-screened: glycine, DL-alanine, beta-alanine, serine, threonine, isoleucine, DL-

valine, piperazine-2-carboxilic acid, proline, arginine, gamma-aminobutyric acid,

ornithine, taurine, creatine and histidine. The biocompatibility of the absorbents was

checked based on values of oral LD50 reported in the safety data sheets of the

compounds. LD50 (lethal dose 50 %) is a measure of the toxicity of a compound which

is an important indicator of its biocompatibility. Glycine, isoleucine, proline, arginine,

ornithine, taurine and histidine present values of oral LD50 in rats between 5000 and

15 000, which means that they are practically non toxic (CCOHS, 2008). No

information about LD50 is available for the other compounds. Concerning hazardous

information, only arginine is irritant. Nine new and non-commercially available

absorbents synthesised by CSIC were also pre-screened using the described

methodology.

All the absorbents studied proved to be able to absorb 2CO at significant rates and

absorption capacities. Nevertheless, the method used was not able to accurately

differentiate them. A more precise and quantitative analysis require time consuming

14

experiments and larger amounts of these substances. Since, unlike alkanolamines, there

is little information in literature about the 2CO absorption in amino acid salt solutions,

the potassium salt of the simplest amino acid (glycine) was selected for this more

extensive characterization. However, the molecular structure of potassium glycinate, a

primary and non-sterically hindered amino acid, makes it expectable to be difficult to

regenerate. Given the considerations made in section 1.3, potassium threonate is likely

to overcome this problem and at the same time to have acceptable reaction kinetics (this

is clarified in Chapter 3). For this reason, this compound was also selected for further

characterization.

The structural formulas of the amino acids glycine and trheonine are presented in Table

1.1.

Table 1.1 - Structural formulas of the amino acids characterized in the present

dissertation.

Glycine Threonine

The present dissertation is organized as follows.

In Part II, the selected amino acids salts are characterized for 2CO absorption purposes:

potassium glycinate (Chapter 2) and potassium threonate (Chapter 3). Physical

properties of their aqueous solutions, including density, viscosity and 2CO physical

solubility are determined at different temperatures and amino acid salt concentrations.

Reaction kinetics towards 2CO is also measured at different temperature and

concentration conditions.

Part III (Chapter 4) reports the study of the 2CO absorption equilibrium in amino acid

salt solutions. Absorption equilibrium in potassium glycinate is measured at different

15

temperatures and amino acid salt concentrations. The equilibrium absorption capacity of

potassium threonate is also determined for one condition of concentration and

temperature and compared to the results for potassium glycinate.

In Part IV (Chapter 5), a bi-dimensional model to describe the membrane contactor

process under study is proposed. The performance of the hollow fiber membrane

contactor using potassium glycinate solutions is analysed (based on the physical,

kinetics and equilibrium data determined in the previous chapters). The influence of the

system parameters on the separation achieved is discussed and a contactor design and a

set of operating conditions are suggested.

Finally, in Part V (Chapter 6), the main conclusions are summarised and suggestions for

future work are presented.

Further details on the experimental setups used along this work are presented in

Appendix A.

1.5. References

Al-Marzouqi, M. H., El-Naas, M. H., Marzouk, S. A. M., Al-Zarooni, M. A., Abdullatif,

N. and Faiz, R. (2008). "Modeling of CO2 absorption in membrane contactors."

Separation and Purification Technology, 59(3), 286-293.

Austgen, D. M., Rochelle, G. T., Peng, X. and Chen, C. C. (1989). "Model of Vapor

Liquid Equilibria for Aqueous Acid Gas Alkanolamine Systems Using the Electrolyte

Nrtl Equation." Industrial & Engineering Chemistry Research, 28(7), 1060-1073.

Baum, J. A. and Woehlck, H. J. (2003). "Interaction of inhalational anaesthetics with

CO2 absorbents " Best Practice and Research Clinical Anaesthesiology, 17(1), 63-76.

Bishnoi, S. and Rochelle, G. T. (2000). "Absorption of carbon dioxide into aqueous

piperazine: reaction kinetics, mass transfer and solubility." Chemical Engineering

Science, 55(22), 5531-5543.

16

Blauwhoff, P. M. M., Versteeg, G. F. and Vanswaaij, W. P. M. (1984). "A Study on the

Reaction between CO2 and Alkanolamines in Aqueous-Solutions." Chemical

Engineering Science, 39(2), 207-225.

Bonenfant, D., Mimeault, M. and Hausler, R. (2003). "Determination of the structural

features of distinct amines important for the absorption of CO2 and regeneration in

aqueous solution." Industrial & Engineering Chemistry Research, 42(14), 3179-3184.

Brandi, L. S. (2008). "Anesthesia - Information for patients -

http://www.brandianestesia.it/." Retrieved October, 2008, 2008.

CAETS (1995). The Role of Technology in Environmentally Sustainable Development.

Kiruna, Sweden, Council of Academies of Engineering and Technological Sciences.

Caplow, M. (1968). "Kinetics of carbamate formation and breakdown." Journal of the

American Chemical Society, 90(24), 6795-6803.

CCOHS. (2008). "What is an LD50 and LC50? - http://www.ccohs.ca." Retrieved

October, 2008, 2008.

Derks, P. W. J., Kleingeld, T., van Aken, C., Hogendoom, J. A. and Versteeg, G. F.

(2006). "Kinetics of absorption of carbon dioxide in aqueous piperazine solutions."

Chemical Engineering Science, 61(20), 6837-6854.

Dingley, J., Ivanova-Stoilova, T. M., Grundler, S. and Wall, T. (1999). "Xenon: recent

developments." Anaesthesia, 54(4), 335-346.

Dosch, M. P. (2004). "The Anesthesia Gas Machine - http://www.udmercy.edu."

Retrieved October, 2008, 2008.

Fan, S. Z., Lin, Y. W., Chang, W. S. and Tang, C. S. (2008). "An evaluation of the

contributions by fresh gas flow rate, carbon dioxide concentration and desflurane partial

pressure to carbon monoxide concentration during low fresh gas flows to a circle

anaesthetic breathing system." European Journal of Anaesthesiology, 25(8), 620-626.

Favre, E. (2007). "Carbon dioxide recovery from post-combustion processes: Can gas

permeation membranes compete with absorption?" Journal of Membrane Science,

294(1-2), 50-59.

17

Feron, P. and Jansen, A. (2002). "CO2 separation with polyolefin membrane contactors

and dedicated absorption liquids: performances and prospects." Separation and

Purification Technology, 27(3), 231-242.

Gabelman, A. and Hwang, S. (1999). "Hollow fiber membrane contactors." Journal of

Membrane Science, 159(1-2), 61-106.

Goff, G. S. and Rochelle, G. T. (2006). "Oxidation inhibitors for copper and iron

catalyzed degradation of monoethanolamine in CO2 capture processes." Industrial &

Engineering Chemistry Research, 45(8), 2513-2521.

Hampe, E. M. and Rudkevich, D. M. (2003). "Exploring reversible reactions between

CO2 and amines." Tetrahedron, 59(48), 9619-9625.

Holst, J., Politiek, P. P., Niederer, J. P. M. and Versteeg, G. F. (2006). "CO2 capture

from flue gas using amino acid salt solutions". GHGT-8, NTNU VIDERE, Pav. A,

Dragvoll, NO-7491 Trondheim, Norway.

Hook, R. J. (1997). "An investigation of some sterically hindered amines as potential

carbon dioxide scrubbing compounds." Industrial & Engineering Chemistry Research,

36(5), 1779-1790.

Idem, R. and Tontiwachwuthikul, P. (2006). "Preface for the special issue on the

capture of carbon dioxide from industrial sources: Technological developments and

future opportunities." Industrial & Engineering Chemistry Research, 45(8), 2413-2413.

Jensen, A., Jensen, J. B. and Faurholt, C. (1952). "Studies on Carbamates .6. The

Carbamate of Glycine." Acta Chemica Scandinavica, 6(3), 395-397.

Keshavarz, P., Ayatollahi, S. and Fathikalajahi, J. (2008). "Mathematical modeling of

gas–liquid membrane contactors using random distribution of fibers." Journal of

Membrane Science, 325, 98–108.

Knolle, E. and Gilly, H. (2000). "Absorption of carbon dioxide by dry soda lime

decreases carbon monoxide formation from isoflurane degradation." Anesthesia and

Analgesia, 91(2), 446-451.

18

Kumar, P., Hogendoorn, J., Feron, P. and Versteeg, G. (2001). "Density, viscosity,

solubility, and diffusivity of N2O in aqueous amino acid salt solutions." Journal of

Chemical and Engineering Data, 46(6), 1357-1361.

Kumar, P., Hogendoorn, J., Feron, P. and Versteeg, G. (2002a). "New absorption liquids

for the removal of CO2 from dilute gas streams using membrane contactors." Chemical


Kumar, P., Hogendoorn, J., Timmer, S., Feron, P. and Versteeg, G. (2003a).

"Equilibrium solubility of CO2 in aqueous potassium taurate solutions: Part 2.

Experimental VLE data and model." Industrial & Engineering Chemistry Research,

42(12), 2841-2852.

Kumar, P., Hogendoorn, J., Versteeg, G. and Feron, P. (2003b). "Kinetics of the

reaction of CO2 with aqueous potassium salt of taurine and glycine." American Institute

of Chemical Engineers Journal, 49(1), 203-213.

Kumar, P. S. (2002). "Development and design of membrane gas absorption processes".

Enschede, University of Twente.

Kumar, P. S., Hogendoorn, J. A., Feron, P. H. M. and Versteeg, G. F. (2002b). "New

absorption liquids for the removal of CO2 from dilute gas streams using membrane

contactors." Chemical Engineering Science, 57(9), 1639-1651.

Kumar, P. S., Hogendoorn, J. A., Feron, P. H. M. and Versteeg, G. F. (2003c).


Crystallization in carbon dioxide loaded aqueous salt solutions of amino acids."

Industrial & Engineering Chemistry Research, 42(12), 2832-2840.

Li, J. L. and Chen, B. H. (2005). "Review Of CO2 absorption using chemical solvents in

hollow fiber membrane contactors." Separation and Purification Technology, 41(2),

109-122.

Ma'mun, S., Svendsen, H. F., Hoff, K. A. and Juliussen, O. (2007). "Selection of new

absorbents for carbon dioxide capture." Energy Conversion and Management, 48(1),

251-258.

19

Majchrowicz, M., Niederer, J. P. M., Velders, A. H. and Versteeg, G. F. (2006).

"Precipitation in amino acid salt CO2 absorption systems". GHGT-8 NTNU VIDERE,

Pav. A, Dragvoll, NO-7491 Trondheim, Norway.

Marini, F., Bellugi, I., Gambi, D., Pacenti, M., Dugheri, S., Focardi, L. and Tulli, G.

(2007). "Compound A, formaldehyde and methanol concentrations during low-flow

sevoflurane anaesthesia: comparison of three carbon dioxide absorbers." Acta

Anaesthesiologica Scandinavica, 51(5), 625-632.

Mendes, A. M. M. (2000). "Development of an adsorption/membrane based system for

carbon dioxide, nitrogen and spur gases removal from a nitrous oxide and xenon

anaesthetic closed loop." Acp-Applied Cardiopulmonary Pathophysiology, 9(2), 156-

163.

Metz, B., Davidson, O., Coninck, H. d., Loos, M. and Meyer, L. (2005). IPCC Special

Report on Carbon Dioxide Capture and Storage. C. U. Press. Cambridge, New York,

Melbourne, Madrid, Cape Town, Singapore, São Paulo, Intergovernmental Panel on

Climate Change.

Murray, J. M., Renfrew, C. W., Bedi, A., McCrystal, C. B., Jones, D. S. and Fee, J. P.

H. (1999). "Amsorb - A new carbon dioxide absorbent for use in anesthetic breathing

systems." Anesthesiology, 91(5), 1342-1348.

Park, J. Y., Yoon, S. J. and Lee, H. (2003). "Effect of steric hindrance on carbon

dioxide absorption into new amine solutions: Thermodynamic and spectroscopic

verification through solubility and NMR analysis." Environmental Science &

Technology, 37(8), 1670-1675.

Penny, D. E. and Ritter, T. J. (1983). "Kinetic-Study of the Reaction between Carbon-

Dioxide and Primary Amines." Journal of the Chemical Society-Faraday Transactions I,

79, 2103-2109.

Pontes, S. L. d. R. (2006). "Removal of carbon dioxide and nitrogen from a xenon based

closed-circuit anaesthetic system". Porto, Porto.

Rangwala, H. A. (1996). "Absorption of carbon dioxide into aqueous solutions using

hollow fiber membrane contactors." Journal of Membrane Science, 112(2), 229-240.

20

Reinelt, H., Marx, T., Schirmer, U. and Schmidt, M. (2001). "Xenon expenditure and

nitrogen accumulation in closed-circuit anaesthesia." Anaesthesia, 56(4), 309-311.

Rochelle, G. T., Bishnoi, S., Chi, S., Dang, H. and Santos, J. (2001). Research needs for

CO2 capture from flue gas by aqueous absorption/stripping. U. S. D. o. E.-F. E. T.

Center. Pittsburg, U. S. Department of Energy - Federal Energy Technology Center: 60.

Santos, J. C., Magalhaes, F. D. and Mendes, A. (2008). "Contamination of zeolites used

in oxygen production by PSA: Effects of water and carbon dioxide." Industrial &


Sartori, G. and Savage, D. W. (1983). "Sterically Hindered Amines for CO2 Removal

from Gases." Industrial & Engineering Chemistry Fundamentals, 22(2), 239-249.

Singh, P., Nierderer, J. P. M. and Versteeg, G. F. (2007). "Structure and activity

relationships for amine based CO2 absorbents - I." International Journal of Greenhouse

Gas Control, 1(1), 5-10.

Struys, M. M. R. F., Bouche, M. P. L. A., Rolly, G., Vandevivere, Y. D. I., Dyzers, D.,

Goeteyn, W., Versichelen, L. F. M., Van Bocxlaer, J. F. P. and Mortier, E. P. (2004).

"Production of compound A and carbon monoxide in circle systems: an in vitro

comparison of two carbon dioxide absorbents." Anaesthesia, 59(6), 584-589.

Supap, T., Idem, R., Tontiwachwuthikul, P. and Saiwan, C. (2006). "Analysis of

monoethanolamine and its oxidative degradation products during CO2 absorption from

flue gases: A comparative study of GC-MS, HPLC-RID, and CE-DAD analytical

techniques and possible optimum combinations." Industrial & Engineering Chemistry

Research, 45(8), 2437-2451.

UNFCCC (2008). United Nations Framework Convention on Climate change,

http://unfccc.int.

Versteeg, G. F., Van Dijck, L. A. J. and Van Swaaij, W. P. M. (1996). "On the kinetics

between CO2 and alkanolamines both in aqueous and non-aqueous solutions. An

overview." Chemical Engineering Communications, 144, 113-158.

21

Versteeg, G. F. and Van Swaaij, W. P. M. (1988). "Solubility and Diffusivity of Acid

Gases (CO2, N2O) in Aqueous Alkanolamine Solutions." Journal of Chemical and

Engineering Data, 33(1), 29-34.

Vladisavljevic, G. T. (1999). "Use of polysulfone hollow fibers for bubbleless

membrane oxygenation/deoxygenation of water." Separation and Purification

Technology, 17(1), 1-10.

Weiland, R. H., Chakravarty, T. and Mather, A. E. (1993). "Solubility of Carbon-

Dioxide and Hydrogen-Sulfide in Aqueous Alkanolamines." Industrial & Engineering

Chemistry Research, 32(7), 1419-1430.

Whalen, F. X., Bacon, D. R. and Smith, H. M. (2005). "Inhaled anesthetics: an historical

overview." Best Practice Research Clinical Anaesthesiology, 19(3), 323-330.

Wickramasinghe, S. R., Han, B., Garcia, J. D. and Specht, R. (2005). "Microporous

membrane blood oxygenators." American Institute of Chemical Engineers Journal,

51(2), 656-670.

Yan, S. P., Fang, M. X., Zhang, W. F., Wang, S. Y., Xu, Z. K., Luo, Z. Y. and Cen, K.

F. (2007). "Experimental study on the separation of CO2 from flue gas using hollow

fiber membrane contactors without wetting." Fuel Processing Technology, 88(5), 501-

511.

Zhang, Q. and Cussler, E. L. (1985a). "Microporous Hollow Fibers for Gas-Absorption

.1. Mass-Transfer in the Liquid." Journal of Membrane Science, 23(3), 321-332.

Zhang, Q. and Cussler, E. L. (1985b). "Microporous Hollow Fibers for Gas-Absorption

.2. Mass-Transfer across the Membrane." Journal of Membrane Science, 23(3), 333-

345.

Part II

25

2. Characterization of potassium glycinate

for carbon dioxide absorption purposes 1

Abstract

Aqueous solutions of potassium glycinate were characterized for carbon dioxide

absorption purposes. Density and viscosity of these solutions, with concentrations

ranging from 0.1 to 3 M, were determined at temperatures from 293 to 313 K.

Diffusivity of 2CO in solution was estimated applying the modified Stokes-Einstein

relation. Solubilities of 2N O at the same temperatures and concentrations were

measured and the ion specific parameter based on the Schumpe’s model was determined

for the glycinate anion; the solubilities of 2CO in these solutions were then computed.

The reaction kinetics of 2CO in the aqueous solution of potassium glycinate was

determined at 293, 298 and 303 K using a stirred cell reactor. The results were

interpreted using the DeCoursey equation for the calculation of the enhancement factor.

The rate of absorption as a function of the temperature and solution concentration for

the conditions studied was found to be given by the following expression:

( )2 2

16 85442.42 10 exp exp 0.44CO S S COr C C C

T

− − = ×

.

1 Portugal, A. F.; Derks, P. W. J.; Versteeg, G. F.; Magalhães, F. D.; Mendes, A., “Characterization of potassium glycinate for carbon dioxide absorption purposes”, Chem. Eng. Sci., 2007, 62(23), 6534 – 6547

26

2.1. Introduction

The carbon dioxide removal from closed anesthetic loops is currently achieved using

soda lime canisters (a mixture of calcium, potassium and sodium hydroxides) which is

an unsafe technique (Mendes, 2000). The use of dehydrated soda lime is associated to

explosions due to the hydrogen formation and excessive heating during the reaction

with carbon dioxide. Soda lime can also originate toxic compounds resulting from the

reaction with some halogenated anesthetics (Whalen et al., 2005). Because of that and

because exhausted soda lime canisters are a hospital solid waste, this outdated system

needs to be replaced by a safer technology. A possible candidate is the use of absorption

membrane contactors. This strategy presents various advantages. The use of a dense

highly permeable membrane isolates the absorption system from the anesthetic loop and

the absorption solution can be regenerated after contacting with carbon dioxide.

However, the absorbent should have a suitable carbon dioxide absorption kinetics and

capacity, negligible vapor pressure, high chemical and thermal stability and should be

harmless to the patient.

Mostly as a consequence of the Kyoto protocol, the new stringent environmental

regulations towards the emission of acidic gases raised concern about carbon dioxide

capture and storage In the last decades, hollow fiber membrane contactors have been

studied using absorbent aqueous solutions such as alkanolamines or blends of

alkanolamines for the selective removal of acid gases like 2H S and 2CO from a variety

of industrial and natural gas streams (Al-Juaied and Rochelle, 2006; Kumar et al.,

2003c). However, for applications in highly oxygenated environments, such as flue gas

treatment, life support systems or anesthetic gas circuits, alkanolamines might not be of

interest since they undergo oxidative degradation (Goff and Rochelle, 2006; Kumar et

al., 2003c; Supap et al., 2006). Amino acids are now being studied as a possible

alternative for alkanolamines (Feron and Jansen, 2002). Although being more

expensive, a few advantages make amino acids attractive like being generally more

stable to oxidative degradation and presenting lower volatilities while showing similar

absorption kinetics and capacities in comparison to alkanolamine solutions (Kumar et

al., 2003b). Moreover, amino acids aqueous solutions have higher surface tensions and

the viscosities are very similar to water’s. If membrane contactors are to be used, it is

27

important to consider that the membrane materials should be compatible with the

absorption liquid. A liquid with higher surface tension and lower corrosiveness will

make possible the efficient use of cheaper and commercially available membranes,

economically improving the process (Kumar et al., 2003a).

Despite of rising interest, few studies have been performed so far on amino acids as

carbon dioxide absorbents. TNO Environment Energy and Process Innovation has been

developing a process for carbon dioxide removal from flue gas process based on the use

of amino acids and salts (Feron and Jansen, 2002). Kumar and co-workers studied in

detail the absorption of carbon dioxide in potassium salts of taurine and briefly analyzed

glycine (Kumar et al., 2001, 2002, 2003a, 2003b, 2003c). Holst et al. (2006) compared

the apparent absorption rate constants of 2CO with different amino acid salt solutions

and concluded that they were comparable with alkanolamines. Recently Lee et al.

studied the physical properties and the absorption kinetics of sodium glycinate as an

absorbent of carbon dioxide (Lee et al., 2005, 2006, 2007; Song et al., 2006). However,

the data available in literature is still too limited to permit a suitable design and

optimization of processes using amino acid absorbents.

After a pre-screening of a set of different amino acid salts, potassium glycinate

presented several interesting properties, such as very good thermal stability and fast

apparent reaction rate towards carbon dioxide. Besides, it is commercially available and

relatively cheap. For these reasons, it was selected for characterization as a carbon

dioxide absorbent in the present work. This includes the determination of the densities

and viscosities of aqueous solutions with concentrations between 0.1 to 3 M and

temperatures from 293 to 313 K. The solubility of 2N O in potassium glycinate

solutions was also measured and the absorption kinetics of carbon dioxide in potassium

glycinate solutions obtained.

2.2. Zwitterion Reaction Mechanism

The zwitterion mechanism, originally proposed by Caplow (1968) is generally applied

to model the carbon dioxide absorption in amino acid aqueous solutions. According to

the zwitterion mechanism, 2CO reacts with the amino acid salt (potassium glycinate in

28

the present case) forming a zwitterion that is subsequently deprotonated by a base

present in solution.

Formation of the potassium glycinate zwitterion

2

12 2 2 2 2

k

kH N CH COO K CO OOC H N CH COO K

−

− + − + − +→− − + − −← (1)

Removal of a proton by a base

2 2 2B

B

k

i ikOOC H N CH COO K B OOCHN CH COO K B H

−

− + − + − − + +→− − + − − +← (2)

where iB are the bases present in solution able to deprotonate the zwitterion. In amino

acid salt solutions, these bases are 2H O , OH − and the amino acid salt

2 2H NCH COO K− + (Blauwhoff et al., 1984).

Assuming quasi steady-state condition for the zwitterion concentration and since the

second proton transfer step can be considered irreversible, the overall reaction rate,

2COr− , can then be obtained:

2 2

2

11i i

CO S CO

B Bi

kr C C

k

k c−

− =+∑

(3)

where SC is the concentration of the amino acid salt and 2COC is the concentration of

carbon dioxide in the liquid. Limiting conditions lead to simplified reaction rate

expressions that are well described in literature (Derks et al., 2006; Kumar et al.,

2003c).

During the absorption, carbon dioxide reacts also with the hydroxide ions present in

solution:

2 3OHkCO OH HCO− −+ → (4)

Taking reaction (4) into account, the overall reaction rate (3) becomes:

29

2 2

2

11i i

CO S COOH OH

B Bi

kr C k C C

k

k C

− −

−

− = + +

∑

(5)

However, as potassium glycinate is a weak base, the contribution of reaction (4) to the

overall reaction kinetics can be considered negligible as well as the contribution of

OH − to the deprotonation of the zwitterion (Kumar et al., 2003c). The overall rate of

reaction of 2CO with potassium glycinate therefore becomes:

( ) ( )

2

2

2 2

2

1 1

11

S COCO

H O H O AmA S

k C Cr

k k C k k C− −

− =+

+

(6)

Primary amines such as monoethanolamine (MEA) usually react with 2CO following a

second order reaction kinetics, which means that the deprotonation of the zwiteterion is

relatively fast when compared to the reversion rate of 2CO and the amine

( 1 1i iB B

i

k

k c− <<

∑). Equation (6) is then reduced to:

2 22CO S COr k C C− = (7)

A thermodynamically sound model for the calculation of the kinetic constant should be

expressed in terms of activities rather than concentrations (Haubrock et al., 2005).

However, such a model would require the knowledge of a number of parameters

including equilibrium data that are not available. To account for the solution non-

idealities it is common to use a semi empirical model which relates the kinetic constant

to the solution ionic strength (Cullinane and Rochelle, 2006):

( )expeffk k bI= (8)

where effk is the effective kinetic constant, corrected for the solution ionic strength, b

is a constant and I is the ionic strength given by 21

2 i iI C z= ∑ , where iC and iz are

respectively the molar concentration and the charge of ion i in solution. This model is

not thermodynamically sound and cannot be extrapolated for different ions present in

solution since all the solution non-idealities are lumped in the effective kinetic constant,

30

effk ; however it is generally sufficient to represent the experimental data of a single

absorption system (Haubrock et al., 2005).

2.3. Mass Transfer

The absorption of a pure gas (carbon dioxide in the present work) into a lean reactive

liquid (potassium glycinate solution) is described by the following equation

(Danckwerts, 1970)

2

2

2

COCO L

CO

PN E k A

H= ⋅ (9)

where 2CON is the molar flow of 2CO entering the liquid, Lk is the physical mass

transfer coefficient, 2COP is the 2CO partial pressure in the gas phase,

2COH is the Henry

constant of 2CO in solution, A is the interfacial area between the gas and the liquid

phases and E is the enhancement factor. The enhancement factor represents the ratio

between the rate of absorption in the presence of the chemical reaction and the physical

rate of absorption. When the reaction rate is sufficiently high, the reaction occurs

entirely in the liquid film and not in the liquid bulk and the absorption rate can be

divided into three main regimes depending on the dimensionless Hatta number:

2ov CO

L

k DHa

k= (10)

where ovk is the overall reaction kinetic constant (2 2ov CO COk r C= − ) and

2COD is the

diffusion coefficient of 2CO in solution.

Fast pseudo-first order (PFO) reaction regime can be assumed if the following criterion

is fulfilled (Danckwerts, 1970):

3 Ha E∞< << (11)

In this case, the processes of diffusion and reaction occur in parallel in the liquid film.

The enhancement factor can be considered equal to the Hatta number and the gas

absorption rate becomes, therefore, independent of the physical mass transfer

coefficient. The infinite enhancement factor, E∞ , corresponds to a situation of

31

instantaneous reaction and can be estimated, according to the penetration theory, by the

following equation (Danckwerts, 1970; Higbie, 1935):

2

2 2

2

2

CO S S

COS COCO

CO

D C DE

PD D

Hν

∞ = + (12)

where SD is the amino acid salt diffusion coefficient and 2COν is the stoichiometric

coefficient. The instantaneous reaction regime can be considered when E Ha∞ << .

Between the limiting situations of fast pseudo-first order and instantaneous reaction

regime, there is the intermediate regime. According to DeCoursey, the enhancement

factor in the intermediate regime can be approximated as a function of the Hatta number

and the infinite enhancement factor (DeCoursey, 1974; Van Swaaij and Versteeg,

1992):

( ) ( )22 4

2 12 1 14 1

E HaHa HaE

E EE∞

∞ ∞∞

⋅= − + + +− −−

(13)

Since carbon dioxide reacts with the amino acid salt solution, the physical properties

such as physical solubility and diffusivity cannot be directly measured and need to be

estimated indirectly by analogy with a non-reactive gas with similar properties.

Typically, 2N O is the gas used for this purpose because it has a very similar molecular

configuration, volume and electronic structure and it does not react with the absorbent

solution (Laddha et al., 1981).

Since the amino acid salt solutions are ionic, the so called 2N O analogy cannot be

directly applied to estimate the solubility of 2CO in these solutions. Schumpe proposed

a model to describe the solubility of gases in ionic solutions, which takes into account

the salting out effect observed in electrolyte solutions (Schumpe, 1993; Weisenberger

and Schumpe, 1996). This model enables a reliable estimation of the solubility of 2CO

in electrolyte solutions.

The diffusion coefficient is usually difficult to accurately determine and requires time

consuming experiments. Many authors studied the dependence of the diffusion

32

coefficient on the temperature and on the concentration of the absorbent solution and

concluded that it can be related to the solution viscosity, η , through a modified Stokes-

Einstein equation (Brilman et al., 2001; Joosten and Danckwerts, 1972; Kumar et al.,

2001; Versteeg and Van Swaaij, 1988).

constantD αη = (14)

where α is a constant that depends on the pair gas/solution.

2.4. Experimental

Chemicals

Since the amino acid exists in solution with the amine group protonated, it is necessary

to make it react with a hydroxide salt to deprotonate the amine group enabling it to react

with carbon dioxide. The potassium glycinate aqueous solutions were prepared by

adding to the amino acid an equimolar amount of potassium hydroxide in a volumetric

flask with distilled and deionised water. The concentrations of the solutions were

verified by a standard potentiometric titration with 1N HCl solution.

Density and Viscosity

The density of the solutions was measured using a commercial density meter (DMA 58,

anton Paar GmbH).

Viscosities of potassium glycinate solutions were determined experimentally using a

standard Cannon-Fenske viscosimeter.

N2O solubility

The procedure adopted to measure the solubility of 2N O in the amino acid salt

solutions is described in detail by Derks et al. (2005) and will only be briefly

summarized here. The set-up used is composed of two vessels with calibrated volumes;

one for storing the nitrous oxide and the other for the absorbent solution which is

magnetically stirred. A known volume of solution is transferred to the absorbent vessel

33

and degassed by applying vacuum. The vapour equilibrium is allowed to be reached at a

given temperature; the vapour pressure, vaporP , is then recorded. The gas vessel is filled

with 2N O . A certain amount of 2N O is allowed to enter the absorbent vessel and the

initial pressure, initP , is recorded. The stirrer is then switched on and the solution

equilibrium is allowed to be established. The final pressure, eqP , is recorded as well as

the temperature, initT . The temperature is then set to a different value, T , with the help

of the thermostatic bath and a new equilibrium state is established. The amount of

absorbed gas is calculated applying the ideal gas law. The Henry coefficient for 2N O ,

2N OH , is then computed from the following equation:

( ) ( ) ( )( ) ( ) ( )2

eq vapour LN O

ginit vapour init eq vapour

init

P T P T VH T

RVP P T P T P T

T T

− = − − −

(15)

where gV and LV are respectively the volume of gas and liquid in the absorbent vessel

and R is the universal gas constant.

The solution vapour pressure at each temperature is estimated by the following relation:

( ) ( )2 2

purevapour H O H OP T x P T= (16)

where 2H Ox is the molar fraction of water in solution. The vapour pressure as a function

of the absolute temperature, ( )2

pureH OP T , is obtained from the Antoine equation (Poling et

al., 2001).

The experimental solubility of 2N O as a function of the temperature is hence obtained

using the same sample. The volume of liquid as a function of temperature and the amino

acid molar fraction are obtained using the density and the mass of solution.

Kinetic Measurements

The experiments were performed in a stirred cell reactor with a smooth gas-liquid

interface, with an interfacial area of 3 26.490 10 m−× , operating batchwise with respect to

the liquid phase and semi-continuously with respect to the gas phase. The set-up and

34

procedure are described in detail by Derks et al. (2006) and will be only briefly

summarized here. The stirred cell reactor is connected to a calibrated gas vessel filled

with pure carbon dioxide by means of a pressure controller (Brooks, model 5866, 0-500

mbar, 0.5 FS precision). A fresh potassium glycinate solution, previously degassed by

applying vacuum, is transferred into the stirred reactor. Subsequently, after the vapour-

liquid equilibrium is attained at a given temperature, the gas phase pressure inside the

stirred reactor is recorded, vapourP . One begins the experiment by letting the carbon

dioxide to flow from the gas vessel into the stirred cell reactor. During the experiment

the pressure inside the stirred cell reactor is kept constant, scP , using the pressure

controller, and the flow of absorbed carbon dioxide is computed following the pressure

decrease inside the gas vessel. A sketch of the unit is presented in Figure 2.1.

Figure 2.1 – Simplified scheme of the experimental set-up.

The flow of absorbed carbon dioxide in the stirred reactor, 2 ,CO scN , is given by equation

(9) and 2COP is the carbon dioxide partial pressure in the stirred cell (

2CO sc vapourP P P= − ).

In the gas vessel, by simply applying the ideal gas law, the flow of 2CO is given by:

2 ,

gv gvCO gv

V dPN

RT dt= (17)

where gvdP

dt is the pressure decrease rate in the gas vessel and gvV is the volume of the

gas vessel.

2 2, ,CO gv CO scN N= (18)

35

If fast pseudo-first order regime is fulfilled, it is possible to determine experimentally

the overall reaction kinetic constant, ovk , knowing 2COH and

2COD . When pseudo-first

order is considered, the carbon dioxide flow into the reactor tank is given by:

2

2

2

,CO scgv gvov CO

CO

PV dPk D A

RT dt H= (19)

However, to decide the operating conditions that lead to fast pseudo-first order reaction

regime, it is necessary to calculate Ha , which implies the previous knowledge of ovk .

For experiments performed at a given temperature, absorbent concentration and stirring

speed, the Hatta number is constant. Changing the partial pressure of carbon dioxide

inside the reactor, one changes the infinite enhancement factor and, consequently, the

ratio between Ha and E∞ , which means that the absorption regime changes. By

lowering the carbon dioxide partial pressure at constant Ha , the ratio between the flow

and the partial pressure of carbon dioxide becomes eventually constant, that is the value

of 2 2CO CON P becomes independent of

2COP . Under these conditions it is very probable

that the pseudo-first order reaction regime is attained. Plotting 2CON as a function of

2COP at the fast pseudo-first order regime, the slope of this curve is related with ovk at a

given temperature and amino acid concentration. Figure 2.2 shows experimental values

of 2CON as a function of the partial pressure of carbon dioxide. The slope of the fitted

line is related with ovk through equation (19).

36

Figure 2.2 - 2CON as a function of

2COP at 298 K for a potassium glycinate concentration

of 0.587 M.

For higher carbon dioxide partial pressures, still at constant Ha , the flow of carbon

dioxide into the liquid (absorption) depends not only on the overall kinetic constant but

also on the diffusivity ratio of carbon dioxide and absorbent, 2CO SD D . For sufficiently

high partial pressures, the instantaneous reaction regime is reached when the

enhancement factor becomes independent on the overall kinetic constant.

2.5. Results and discussion

Densities of potassium glycinate aqueous solutions with concentrations from 0.1 to 3 M

and temperatures from 273 to 313 K were determined and are presented in Table 2.1.

37

Table 2.2 – Densities of potassium glycinate solutions - ( )-3kg mρ ⋅

( )KT

( )MSC 293 298 303 313

0 998.29 997.13 995.71 992.25 0.102 1004.37 1003.06 1001.59 997.28 0.296 1015.97 1014.56 1013.00 1001.77 0.594 1033.28 1031.69 1030.02 1025.15 1.003 1056.57 1054.81 1052.98 1047.94 1.992 1112.29 1110.13 1108.01 1102.34 2.984 1163.85 1161.44 1159.07 1150.37

The experimental solubility of 2N O in potassium glycinate solutions is given in Table

2.2 and Figure 2.3.

Table 3.2 - Experimental Henry constants of 2N O in potassium glycinate solutions.

( )MSC ( )KT ( )2

3 -1Pa m molN OH ( )MSC ( )KT ( )2

3 -1Pa m molN OH

0.102 293.2 3640 1.003 293.9 4718 0.102 297.4 4086 1.003 298.3 5368 0.102 298.5 4196 1.003 298.7 5301 0.102 303.1 4719 1.003 303.0 5991 0.296 293.1 3908 1.003 303.4 5854 0.296 293.3 3861 1.003 312.4 7252 0.296 293.4 3876 1.992 293.1 6017 0.296 293.5 3830 1.992 293.2 6140 0.296 298.0 4369 1.992 293.6 6112 0.296 298.2 4455 1.992 293.8 6024 0.296 303.3 4995 1.992 298.1 6711 0.296 311.9 6094 1.992 298.4 6974 0.594 293.4 4241 1.992 302.9 7753 0.594 293.7 4235 1.992 303.0 7519 0.594 298.3 4856 1.992 312.0 9315 0.594 302.2 5375 2.984 293.2 7694 0.594 312.0 6782 2.984 293.3 7621 1.003 293.0 4683 2.984 298.5 8545 1.003 293.1 4539 2.984 303.1 9305 1.003 293.3 4626

38

Figure 2.3 – Experimental Henry constants of 2N O in water and in potassium glycinate

solutions as a function of temperature. Comparison with the solubility in water

determined by Versteeg and Van Swaaij (1988).

The same experimental method was also used to obtain the solubility of 2N O in water

and results compared with the ones by Versteeg and Van Swaaij (1988). It was verified

that they agree within 2% relative error.

The solubility data of 2N O in potassium glycinate was fitted using the Sechenov

relation:

2

2 ,

log N Os

N O w

HK C

H

= ⋅

(20)

where 2N OH and

2 ,N O wH are respectively the Henry constants of 2N O in the amino acid

salt solution and in water. For each concentration and temperature, averaged values of

2N OH were used. For each temperature, the computed Sechenov constants, K , failing

the t-test were rejected based on a 5% confidence limit.

For a single salt, Weisenberger and Schumpe (1996) proposed the following model to

predict the Sechenov’s constant, K :

39

( )i G iK h h n= +∑ (21)

where ih and Gh are the ion and gas specific parameters and in is the valency number

of the ion. In the present work, 1Gly

n − = and 1K

n + = and the Sechenov constant

becomes:

( ) ( )2 2 2N O N O N OGly K

K h h h h− += + + + (22)

Values of parameters K

h + and 2N Oh for the cation and the gas respectively, are reported

in literature (Weisenberger and Schumpe, 1996). These values, together with the

experimental Sechenov’s constant, 2N OK , were used to calculate the anion specific

parameter, Gly

h − . The values of Sechenov’s constant as well as the specific parameters

of gas and cation and the calculated value of the anion parameter are given in Table 2.4.

Table 2.4 – Sechenov’s constants for solubility of 2N O in aqueous potassium glycinate

solutions.

( )KT ( )2

3 -1dm molN OK ( )2

3 -1dm molN Oh ( )3 -1dm molK

h + ( )3 -1dm molGly

h −

293 0.115 -0.0061 0.0922 0.0352 298 0.116 -0.0085 0.0922 0.0408 303 0.112 -0.0109 0.0922 0.0417 313 0.102 -0.0157 0.0922 0.0409

Parameters K

h + and Gly

h − are expected to be essentially constant with the temperature

(Weisenberger and Schumpe, 1996). The values obtained for Gly

h − show, however,

slight temperature dependence. The average Gly

h − over the temperature range is 0.0397.

Taking the values of solubility in water reported by Versteeg and Van Swaaij (1988),

the anion parameter Gly

h − obtained by Kumar et al. (2001) is 0.0413 at 295 K which is

in agreement with the present work. The value of the anion specific parameter Gly

h −

along with the 2CO specific parameter, 2COh , determined by Weisenberger and

Schumpe (1996) enables to predict the Sechenov’s constant of 2CO in potassium

glycinate solutions, 2COK , and subsequently its physical solubility.

40

Table 2.5 – Sechenov’s constants for solubility of 2CO in aqueous potassium glycinate

solutions.

( )KT ( )2

3 -1dm molCOh ( )2

3 -1dm molCOK

293 -0.0155 0.101 298 -0.0172 0.097 303 -0.0189 0.094 313 -0.0223 0.087

The computed physical solubility of carbon dioxide in potassium glycinate solutions is

given in Table 2.6.

Table 2.6 – Henry constants of 2CO in potassium glycinate solutions computed based

on the Sechenov’s model - ( )2

3 -1Pa m molCOH .

( )KT

( )MSC 293 298 303 313

0.10 2710 3044 3405 4217 0.30 2839 3183 3556 4390 0.59 3036 3397 3787 4653 1.0 3340 3725 4138 5053 2.0 4212 4662 5139 6179 3.0 5313 5835 6382 7554

Viscosities of potassium glycinate solutions were determined experimentally and the

Stokes-Einstein relation was used to estimate the diffusion coefficient of 2N O .

Versteeg and Van Swaaij (1988) obtained parameter α of the Stokes-Einstein relation

from experimental values of the diffusion coefficient of 2N O in several alkanolamines

aqueous solutions. These authors proposed 0.8α = , while Brilman et al. (2001)

concluded that the ionic strength of the salt solutions does not influence the diffusion

coefficient. For these reasons, in the present work it is assumed 0.8α = for estimating

of the diffusion coefficient of 2N O in potassium glycinate solutions. Kumar et al.

(2001) studied the diffusivity of 2N O in amino acid salts aqueous solutions and found

0.74α = for potassium taurate. The differences in the calculated diffusivities using one

or the other value for α are lower than 5%, which is within the general experimental

error for the determination of diffusion coefficients.

41

The diffusion coefficient of 2CO in solution is determined using the so called

2 2:N O CO analogy (Gubbins et al., 1966):

2 2

2 2, ,

N O CO

N O w CO w

D D

D D= (23)

The values of the diffusivity of nitrous oxide and carbon dioxide in water were obtained

from the literature (Versteeg and Van Swaaij, 1988). The results of the experimentally

determined viscosities are given in Table 2.7 along with calculated diffusion

coefficients of 2N O and 2CO .

Table 2.7 – Viscosity and diffusivity of 2N O and 2CO in potassium glycinate solutions.

( )MSC ( )KT ( )3 1 110 kg m sη − −× ⋅ ⋅ ( )2

9 2 -110 m sN OD × ⋅ ( )2

9 2 -110 m sCOD × ⋅

293 1.030 1.52 1.67 298 0.914 1.75 1.89 303 0.819 1.99 2.12

0.102

313 0.666 2.57 2.66

293 1.075 1.47 1.61 298 0.962 1.68 1.81 303 0.851 1.93 2.06

0.296

313 0.693 2.49 2.58

293 1.148 1.40 1.53 298 1.020 1.60 1.73 303 0.909 1.8 1.95

0.594

313 0.746 2.35 2.43

293 1.263 1.30 1.42 298 1.136 1.47 1.58 303 1.008 1.69 1.80

1.003

313 0.826 2.16 2.24

293 1.620 1.06 1.16 298 1.449 1.21 1.30 303 1.287 1.39 1.48

1.992

313 1.070 1.76 1.82

293 2.109 0.86 0.94 298 1.861 0.99 1.07 303 1.677 1.12 1.20

2.984

313 1.363 1.45 1.50

42

Overall Kinetic Constants

The overall kinetic constants of the carbon dioxide absorption in potassium glycinate

aqueous solutions were calculated using the described methodology for a potassium

glycinate concentration from 0.1 to 3 M and at 293, 298 and 303 K.

Table 2.8 shows the overall kinetic constants of carbon dioxide absorption obtained at

the potassium glycinate concentrations and temperatures employed. Only the

experimental values of 2CON as a function of

2COP considered to be in the fast pseudo-

first reaction order regime were used for that calculation. The complete set of kinetic

results is shown in Appendix.

Table 2.8 – Experimental values of the overall kinetic constant assuming pseudo-first

order behaviour.

( )7 -1 -1Slope 10 mol mbar s⋅ ⋅ ⋅ ( )-1sovk

( )KT

( )MSC 293 298 303 293 298 303

0.0994 --- 2.51 --- --- 732 --- 0.299 4.44 4.38 5.19 2340 2540 3930 0.587 4.30 5.94 7.37 2640 5590 9490 0.999 7.09 8.83 9.40 9390 16200 20000 1.984 7.92 9.55 12.7 22800 36100 68000 3.005 --- 10.7 --- --- 86300 ---

One must now verify if inequality (11), corresponding to the pseudo-first order reaction

criterion, is fulfilled. The Hatta numbers and the infinite enhancement factors were then

calculated for each experimental condition. However, to calculate the Hatta number one

needs to determine the physical mass transfer coefficient, Lk - see equation (10), and to

calculate the infinite enhancement factors one needs to determine the diffusion

coefficient of potassium glycinate in solution, SD - see equation (12).

The physical mass transfer coefficient, Lk , was calculated using the empirical

expression referred by Versteeg et al. (1987):

3 42Sh Re Scc cc= (24)

43

where Sh, Re and Sc are respectively the Sherwood, Reynolds and Schmidt

dimensionless numbers defined as:

2

Sh L S

CO

k d

D

⋅= (25)

( )2

Re Sd Nρη

= (26)

2

ScCOD

ηρ

=⋅

(27)

where Sd and N are respectively the characteristic dimension and the speed of the

stirrer which are, in the present case, 29.09 10 mSd −= × and 1108minN −= . The

constants 2c , 3c and 4c were determined experimentally performing absorption

experiments of 2CO in water at different temperatures and they show to be within the

usual values for this kind of systems (Versteeg et al., 1987).

0.7279 0.4076Sh 0.1018 Re Sc= ⋅ (28)

It was verified that all computed Ha were much higher than 3 and therefore the first

part of the inequality (11) is confirmed.

The diffusion coefficient of potassium glycinate in solution, SD , was computed

assuming that it follows the modified Stokes-Einstein relation (14) with 0.6α =

(Versteeg and Van Swaaij, 1988). To estimate the diffusion coefficient of the salt at

infinite dilution, 0SD , the Nernst equation for the diffusion in electrolyte solutions was

applied (Poling et al., 2001):

( ) ( )( ) ( )

0

2 0 0

1 1

1 1S

RT z zD

F λ λ+ −

+ −

+ = +

(29)

where F is the Faraday constant, z+ and z− are the valencies of the cation and anion

respectively and 0λ+ and 0λ− are the ionic conductances of the cation and anion

respectively at infinite dilution. Values of 0λ+ at each temperature was calculated based

on the work of Fell and Hutchiso (1971). 0λ− at 298 K was obtained from Miyamoto and

Schmidt (1933) and it was assumed that it depends on the temperature in the same way

44

as 0λ+ . The computed diffusion coefficient of potassium glycinate in solutions, SD , are

shown in Table 2.9.

Table 2.9 – Computed values of SD used to calculate E∞ - ( )10 2 -110 m sSD × ⋅ .

( )KT

( )MSC 293 298 303

0.0994 --- 11.2 --- 0.299 8.56 10.8 13.3 0.587 8.23 10.5 12.8 0.999 7.77 9.80 12.0 1.984 6.69 8.47 10.4 3.005 --- 7.29 ---

The Hatta number, Ha , along with the minimum value of E∞ (corresponding to the

higher pressure used for computing ovk assuming the pseudo-first order) are given in

Table 2.10 for the absorbent concentrations and temperatures studied.

Table 2.10 - Ha and minimum values of E∞ used for computing ovk assuming PFO.

Ha E∞

( )KT

( )MSC 293 298 303 293 298 303

0.0994 --- 38.5 --- --- 182 --- 0.299 73.1 72.9 86.2 316 477 713 0.587 79.4 110 137 437 500 776 0.999 154 194 205 748 937 1460 1.984 261 314 410 1930 2430 3710 3.005 --- 528 ---- --- 4400 ---

Fast pseudo-first order regime can only be ensured for ratios between E∞ and Ha close

to 10. For this reason, the DeCoursey relation was applied and new values for ovk were

calculated by minimizing the sum of the square residues between the experimental

carbon dioxide absorption flow and the flow calculated by applying the DeCoursey

equation. These values are given in Table 2.11.

45

Table 2.11 - Experimental values of the overall kinetic constants of potassium glycinate

calculated using the DeCoursey equation - ( )-1sovk .

( )KT

( )MSC 293 298 303

0.0994 --- 881 --- 0.299 2860 2710 4360 0.587 3130 6420 11000 0.999 11500 19600 22000 1.984 24200 38600 69800 3.005 --- 93900 ---

The deviation of the enhancement factor determined experimentally and the one

calculated based on the DeCoursey equation is presented in Figure 2.4.

Figure 2.4 – Parity plot of experimental enhancement factor and the DeCoursey

approximation.

Since all the experiments were performed at very low loadings, the only ions

contributing to the ionic strength of the solutions are potassium cation and glycinate

anion, both monovalent species (2 1z = ) and thus SI C= . Combining equation (7) with

(8), the absorption rate of 2CO becomes:

( )2 22 expCO S S COr k C bC C− = (30)

46

In addition, assuming that the kinetic constant obeys the Arrhenius equation:

2 2,0 expA

k kT = ⋅

(31)

Plotting ovk as a function of SC it is possible to perform a global fitting to the

experimental results for all temperatures and concentrations considered, in order to

obtain the kinetic parameters 2,0k , A and b . The resulting expression for computation

of the overall kinetic constant as a function of temperature and amino acid salt

concentration, obtained by minimizing the sum of the relative residues squared is (with

SC expressed in mol dm-3):

( )16 -185442.42 10 exp exp 0.44 sov S Sk C C

T

− = ×

(32)

where a coefficient of determination of 0.991 was obtained.

The zwitterion mechanism constants, 2k , ( )2 1H Ok k− and ( )1AmAk k− , were also fitted

using the same procedure but not accounting to the solution ionic strength:

13 -1 -12

58002.81 10 exp M sk

T

− = × ⋅

; ( )2

-1 -11

12651.05 10 exp MH Ok k

T−− = ×

and

( ) 6 -11

53074.89 10 exp MAmAk k

T−− = ×

, where a coefficient of determination of 0.956

was obtained. Both fittings are shown in Figures 2.5 and 2.6. Although the first model

fits better the experimental results (smaller sum of squared residues), generally both

models are in agreement with the experimental results for concentrations up to 3 M. It is

however noticeable that above this concentration it is no longer possible to neglect the

non-idealities of the solution. The second model has 6 fitting parameters while the first

has just 3. The second model has a large number of fitting parameters and over fitting

can easily occur. It is very difficult in such circumstances to identify if the non-idealities

play or not a significant role. The simpler first model becomes then more attractive in

the present work.

47

Figure 2.5 – Overall absorption kinetic constant as a function of potassium glycinate

concentration and for different temperatures: experimental values and model lines.

Solid lines correspond to the model that takes into account the ionic strength and dashed

lines to the zwitterion model.

Figure 2.6 - Apparent absorption kinetic constants as a function of potassium glycinate

concentration and at different temperatures: experimental values and model lines. Solid

lines correspond to the model that takes into account the ionic strength and dashed lines

to the zwitterion model.

48

The values of ovk obtained are in good agreement with the work of Kumar et al. (2003c)

for low concentrations but deviate for higher concentrations. This is possibly due to the

effect of the ionic strength not being taken into account in Kumar’s work. On the other

hand, the results of the present work are quite different from the ones by Lee et al.

(2007). Those authors mention apparent kinetic constants for carbon dioxide absorption

in aqueous sodium glycinate solutions about two orders of magnitude lower. However,

a careful analysis of that work shows that the kinetic measurements were performed far

from the fast pseudo-first order reaction regime, since in some cases E∞ was even lower

than Ha .

It is common to relate the kinetic constant of reaction, 2k , with the apK of the

aminoacid salt by means of a Brønsted plot. Penny and Ritter (1983) determined the

kinetic constant and the apK of several amino acids (including glycinate anion) and

alkanolamines. The values of 2k determined in the present work deviate less than 30%

(relative error) from the Brønsted plot based on the work from Penny and Ritter (1983)

in the entire temperature range. Figure 2.7 presents the Brønsted plot of Penny and

Ritter (1983) along with the results from this work.

Figure 2.7 - Brønsted plot of Penny and Ritter (1983) at 293, 298 and 303 K –

Comparison with the present work.

49

2.6. Conclusions

Density and viscosity of potassium glycinate aqueous solutions ranging from 0.1 and 3

M and at temperatures between 273 and 313 K were obtained. The diffusion coefficient

of 2N O in solution was estimated using a modified Stokes-Einstein relation and the

2CO diffusion coefficient in solution was estimated using the so called 2 2:N O CO

analogy (Gubbins et al., 1966).

The solubility of 2N O in the potassium glycinate solutions was experimentally

determined. The salting out effect of the salt concentration in the solubility showed to

be well described by the Sechenov equation. The specific parameter of the glycinate

anion, based on the Schumpe model (Schumpe, 1993), was calculated

( -30.0397 mol dmGly

h − = ) and the solubility of 2CO in solution was then estimated.

The rate of reaction of 2CO with potassium glycinate was determined in a stirred cell

reactor operating in semi-continuous mode. Two approaches were used to obtain the

relevant parameters of the model. Since the conditions for fast pseudo-first order

reaction regime were apparently not fulfilled, the DeCoursey equation was employed to

calculate the enhancement factor. The results indicate that the reaction kinetics

significantly depend on the ionic strength of the solution. The apparent rate of reaction

is in line with the Brønsted plot based on the work from Penny and Ritter (1983). The

obtained overall kinetic constants point out that potassium glycinate is a fast absorbent

when compared with other absorbents, namely with MEA, which shows an overall

kinetic constant at 298 K for a 1 M solution of 5920 -1s (Glasscock et al., 1991) against

the value of 13400 -1s obtained in the present work for potassium glycinate at the same

concentration and temperature conditions. In the future, these results will be applied in

the design and optimization of a membrane contactor to be used for carbon dioxide

removal from anesthetic gas streams.

2.7. Nomenclature

A Gas-liquid interfacial area, m2

50

C Concentration, M or -3mol m⋅

D Diffusion coefficient, 2 -1m s⋅

Sd Stirrer diameter, m

E Enhancement factor, dimensionless

E∞ Infinite enhancement factor, dimensionless

F Faraday constant, 96500 -1C mol⋅

Ha Hatta number, dimensionless

h Ion and gas specific constants in the Shumpe equation, 3 -1m mol⋅

H Henry coefficient, -3Pa mol m⋅ ⋅

I Ionic strength of the solution, -3mol dm⋅

2COJ Carbon dioxide absorption flux, -2 -1mol m s⋅ ⋅

K Sechenov constant, 3 -1dm mol⋅

1k− Zwitterion kinetic constant of the reverse reaction, s-1

2k Zwitterion kinetic constant of the reaction, M-1 ⋅s-1

AmAk Zwitterion deprotonation rate constant for amino acid, M-1 ⋅s-1

appk Apparent rate constant defined as: app ov Sk k C= , M-1 ⋅s-1

iBk Zwitterion mechanism deprotonation rate constant by base, M-1 ⋅s-1

2H Ok Zwitterion mechanism deprotonation rate constant for water, M-1 ⋅s-1

Lk Liquid phase physical mass transfer coefficient, -1m s⋅

OHk − Reaction rate constant with hydroxide ion M

-1 ⋅s-1

ovk Overall kinetic constant, s-1

N Stirrer speed, rps

2CON Carbon dioxide absorption flow, -1mol s⋅

in Valency number of the ions

2COP Carbon dioxide partial pressure, Pa

2COr− Rate of reaction, -3 -1mol m s⋅ ⋅

R Universal gas constant, 8.314 -1 -1J mol K⋅ ⋅

Re Reynolds number, dimensionless

51

Sh Sherwood number, dimensionless

Sc Schmidt number, dimensionless

T Temperature, K

V Volume, m3

x Molar fraction, -1mol mol⋅

,z z+ − Valencies of the cation and anion

Greek symbols

α Constant from the modified Stokes-Einstein equation

máxα Maximum loading achieved in one experiment, mol

CO2⋅ mol

S-1

Sν Stoichiometric coefficient

η Solution viscosity, -1 -1kg m s⋅ ⋅

ρ Solution density, -3kg m⋅ 0λ+ ,

0λ− Ionic conductances of the cation and anion at infinite dilution, cm2 ⋅ Ω−1

Subscripts

2CO Carbon dioxide

eff Effective (after correcting for the ionic strength)

eq Equilibrium

final Final

g Gas phase

Gly− Glycinate anion

gv Gas vessel

K + Potassium cation

init Initial

L Liquid phase

MEA Monoethanolamine

2N O Nitrous oxide

S Amino acid salt

sc Stirred cell

w Water

52

2.8. References

Al-Juaied, M. and Rochelle, G. T. (2006). "Absorption of CO2 in aqueous

diglycolamine." Industrial & Engineering Chemistry Research, 45(8), 2473-2482.

Blauwhoff, P. M. M., Versteeg, G. F., et al. (1984). "A Study on the Reaction between

CO2 and Alkanolamines in Aqueous-Solutions." Chemical Engineering Science, 39(2),

207-225.

Brilman, D. W. F., Van Swaaij, W. P. M., et al. (2001). "Diffusion coefficient and

solubility of isobutene and trans-2-butene in aqueous sulfuric acid solutions." Journal of




Cullinane, J. T. and Rochelle, G. T. (2006). "Kinetics of carbon dioxide absorption into

aqueous potassium carbonate and piperazine." Industrial & Engineering Chemistry

Research, 45(8), 2531-2545.

Danckwerts, P. (1970). "Gas-Liquid Reactions", McGraw-Hill Book Company.

Decoursey W. J. (1974). "Absorption with Chemical-Reaction - Development of a New

Relation for Danckwerts Model." Chemical Engineering Science, 29(9), 1867-1872.

Derks, P., Hogendoorn, K., et al. (2005). "Solubility of N2O in and density, viscosity,

and surface tension of aqueous piperazine solutions." Journal of Chemical and


Derks, P. W. J., Kleingeld, T., et al. (2006). "Kinetics of absorption of carbon dioxide in

aqueous piperazine solutions." Chemical Engineering Science, 61(20), 6837-6854.

Fell, C. J. D. and Hutchiso, H. P. (1971). "Diffusion Coefficients for Sodium and

Potassium Chlorides in Water at Elevated Temperatures." Journal of Chemical and


53




Glasscock, D. A., Critchfield, J. E., et al. (1991). "CO2 Absorption desorption in

mixtures of methyldiethanolamine with monoethanolamine or diethanolamine."





Gubbins, K. E., Bhatia, K. K., et al. (1966). "Diffusion of Gases in Electrolytic

Solutions." American Institute of Chemical Engineers Journal, 12(3), 548-552.

Haubrock, J., Hogendoorn, J. A., et al. (2005). "The applicability of activities in kinetic

expressions: a more fundamental approach to represent the kinetics of the system CO2-

OH- in terms of activities." International Journal of Chemical Reactor Engineering 3.

Higbie, R. (1935). "The rate of absorption of a pure gas into a still liquid during a short

time of exposure." Transactions of the American Institute of Chemical Engineers, 31,

365–389.

Holst, J., Politiek, P., et al. (2006). "CO2 capture from flue gas using amino acid salt

solutions" GHGT-8 Proceedings.

Joosten, G. E. H. and Danckwerts, P. V. (1972). "Solubility and diffusivity of nitrous-

oxide in equimolar potassium carbonate - potassium bicarbonate solutions at 25 ºC and

1 atm." Journal of Chemical and Engineering Data, 17(4), 452-454.

Kumar, P., Hogendoorn, J., et al. (2001). "Density, viscosity, solubility, and diffusivity

of N2O in aqueous amino acid salt solutions." Journal of Chemical and Engineering

Data, 46(6), 1357-1361.

Kumar, P., Hogendoorn, J., et al. (2002). "New absorption liquids for the removal of

CO2 from dilute gas streams using membrane contactors." Chemical Engineering

Science, 57(9), 1639-1651.

54

Kumar, P., Hogendoorn, J., et al. (2003a). "Equilibrium solubility of CO2 in aqueous

potassium taurate solutions: Part 1. Crystallization in carbon dioxide loaded aqueous

salt solutions of amino acids." Industrial & Engineering Chemistry Research, 42(12),

2832-2840.

Kumar, P., Hogendoorn, J., et al. (2003b). "Equilibrium solubility of CO2 in aqueous

potassium taurate solutions: Part 2. Experimental VLE data and model." Industrial &


Kumar, P., Hogendoorn, J., et al. (2003c). "Kinetics of the reaction of CO2 with aqueous

potassium salt of taurine and glycine." American Institute of Chemical Engineers

Journal, 49(1), 203-213.

Laddha, S. S., Diaz, J. M., et al. (1981). "The N2O Analogy - the Solubilities of CO2 and

N2O in Aqueous-Solutions of Organic-Compounds." Chemical Engineering Science,

36(1), 228-229.

Lee, S., Choi, S., et al. (2005). "Physical properties of aqueous sodium glycinate

solution as an absorbent for carbon dioxide removal." Journal of Chemical and


Lee, S., Song, H. J., et al. (2007). "Kinetics of CO2 absorption in aqueous sodium

glycinate solutions." Industrial & Engineering Chemistry Research, 46(5), 1578-1583.

Lee, S., Song, H. J., et al. (2006). "Physical solubility and diffusivity of N2O and CO2 in

aqueous sodium glycinate solutions." Journal of Chemical and Engineering Data, 51(2),

504-509.




163.

Miyamoto, S. and Schmidt, C. L. A. (1933). "Transference and conductivity studies on

solutions of certain proteins and amino acids with special reference to the formation of

complex ions between the alkaline earth elements and certain proteins." Journal of

Biological Chemistry, 99(2), 335.

55


Dioxide and Primary Amines." Journal of the Chemical Society-Faraday Transactions I,

79, 2103-2109.

Poling, B. E., Prausnitz, J. M., et al. (2001). "The Properties of Gases and Liquids".

McGraw-Hill Book Company

Schumpe, A. (1993). "The Estimation of Gas Solubilities in Salt-Solutions." Chemical


Song, H. J., Lee, S., et al. (2006). "Solubilities of carbon dioxide in aqueous solutions of

sodium glycinate." Fluid Phase Equilibria, 246(1-2), 1-5.

Supap, T., Idem, R., et al. (2006). "Analysis of monoethanolamine and its oxidative

degradation products during CO2 absorption from flue gases: A comparative study of

GC-MS, HPLC-RID, and CE-DAD analytical techniques and possible optimum

combinations." Industrial & Engineering Chemistry Research, 45(8), 2437-2451.

Van Swaaij, W. P. M. and Versteeg, G. F. (1992). "Mass-Transfer Accompanied with

Complex Reversible Chemical-Reactions in Gas-Liquid Systems - an Overview."

Chemical Engineering Science, 47(13-14), 3181-3195.

Versteeg, G. F., Blauwhoff, P. M. M., et al. (1987). "The Effect of Diffusivity on Gas-

Liquid Mass-Transfer in Stirred Vessels - Experiments at Atmospheric and Elevated

Pressures." Chemical Engineering Science, 42(5), 1103-1119.




Weisenberger, S. and Schumpe, A. (1996). "Estimation of gas solubilities in salt

solutions at temperatures from 273 K to 363 K." American Institute of Chemical

Engineers Journal, 42(1), 298-300.

Whalen, F. X., Bacon, D. R., et al. (2005). "Inhaled anesthetics: an historical overview."

Best Practice Research Clinical Anaesthesiology, 19(3), 323-330.

56

2.A. Experimental kinetic data

Values of the carbon dioxide flux, 2COJ , as a function of the carbon dioxide partial

pressure for all temperatures and potassium glycinate concentrations studied are

presented in Tables 2.A1-2.A14.

All the experiments began with fresh solution. The maximum loading reached at the end

of each experiment, máxα , is also shown.

The values used to calculate ovk considering pseudo first order reaction regime and

using the DeCoursey approach are marked respectively with PFO and DeCo. Since at high

carbon dioxide partial pressures, the overall kinetic constant plays a minor role on the

enhancement factor and the values of 2COD and SD are estimated, only experiments at

low partial pressures were used to calculate ovk even when the DeCoursey approach

was used.

57

Table 2.A1 – Kinetic data of the reaction of 2CO with potassium glycinate at 0.0994 M and 298 K.

( )2

210

Pa

COP −× ( )

2

4

-1 -2

10

mol s m

COJ ×

⋅ ⋅ ( )

2

-1CO AmAmol mol

máxα

⋅

( )2

210

Pa

COP −× ( )

2

4

-1 -2

10

mol s m

COJ ×

⋅ ⋅ ( )

2

-1CO AmAmol mol

máxα

⋅

3.42 PFO/DeCo 1.34 0.010 16.50 DeCo 4.78 0.037 5.05 PFO/DeCo 2.05 0.015 21.65 DeCo 5.31 0.042 6.43 PFO/DeCo 2.40 0.017 31.54 DeCo 6.52 0.054 11.44 DeCo 3.94 0.029 52.11 DeCo 7.65 0.068

Table 2.A2 - Kinetic data of the reaction of 2CO with potassium glycinate at 0.299 M and 293 K

( )2

210

Pa

COP −×

( )2

4

-1 -2

10

mol s m

COJ ×

⋅ ⋅ ( )

2

-1CO AmAmol mol

máxα

⋅

( )2

210

Pa

COP −×

( )2

4

-1 -2

10

mol s m

COJ ×

⋅ ⋅ ( )

2

-1CO AmAmol mol

máxα

⋅

4.71 PFO/DeCo 3.36 0.008 12.57 DeCo 8.35 0.019 5.66 PFO/DeCo 4.27 0.010 14.60 DeCo 8.49 0.020 6.27 PFO/DeCo 4.57 0.011 19.84 10.0 0.023 6.76 PFO/DeCo 4.92 0.012 24.81 12.4 0.019 7.28 PFO/DeCo 5.58 0.013 34.73 15.2 0.023 8.28 PFO/DeCo 5.34 0.013 40.05 16.2 0.038 8.85 PFO/DeCo 5.48 0.013 44.87 16.1 0.039 9.84 PFO/DeCo 6.38 0.015 54.97 16.8 0.023

57

58


( )2

210

Pa

COP −× ( )

2

4

-1 -2

10

mol s m

COJ ×

⋅ ⋅ ( )

2

-1CO AmAmol mol

máxα

⋅

( )2

210

Pa

COP −× ( )

2

4

-1 -2

10

mol s m

COJ ×

⋅ ⋅ ( )

2

-1CO AmAmol mol

máxα

⋅

2.71 PFO/DeCo 1.78 0.004 9.76 DeCo 5.80 0.013 3.52 PFO/DeCo 2.50 0.002 9.82 DeCo 6.27 0.015 4.77 PFO/DeCo 3.09 0.007 10.56 DeCo 6.44 0.015 5.89 PFO/DeCo 4.14 0.010 11.66 DeCo 6.79 0.016 6.71 PFO/DeCo 4.33 0.010 16.71 9.02 0.021 7.18 PFO/DeCo 4.99 0.012 26.66 13.3 0.031 7.74 PFO/DeCo 5.16 0.012 41.96 16.3 0.040

8.82 DeCo 5.55 0.013


( )2

210

Pa

COP −×

( )2

4

-1 -2

10

mol s m

COJ ×

⋅ ⋅ ( )

2

-1CO AmAmol mol

máxα

⋅

( )2

210

Pa

COP −×

( )2

4

-1 -2

10

mol s m

COJ ×

⋅ ⋅ ( )

2

-1CO AmAmol mol

máxα

⋅

2.10 PFO/DeCo 1.64 0.004 11.34 DeCo 8.62 0.021 3.94 PFO/DeCo 3.12 0.007 16.37 11.6 0.027 5.22 PFO/DeCo 4.36 0.010 31.74 17.3 0.040 6.00 PFO/DeCo 4.66 0.010 56.62 21.2 0.051

8.17 DeCo 6.62 0.015

58

59


( )2

210

Pa

COP −×

( )2

4

-1 -2

10

mol s m

COJ ×

⋅ ⋅ ( )

2

-1CO AmAmol mol

máxα

⋅

( )2

210

Pa

COP −×

( )2

4

-1 -2

10

mol s m

COJ ×

⋅ ⋅ ( )

2

-1CO AmAmol mol

máxα

⋅

2.16 PFO/DeCo 1.54 0.002 12.19 PFO/DeCo 8.11 0.010 2.22 PFO/DeCo 1.50 0.002 15.00 PFO/DeCo 9.67 0.007 4.43 PFO/DeCo 3.43 0.004 15.00 PFO/DeCo 9.73 0.008 4.46 PFO/DeCo 3.29 0.004 23.75 13.5 0.013 7.62 PFO/DeCo 5.28 0.006 24.79 13.5 0.011 9.47 PFO/DeCo 6.26 0.006 24.85 13.5 0.013 9.51 PFO/DeCo 6.38 0.009 35.42 19.4 0.015 10.12 PFO/DeCo 6.62 0.009 45.50 22.7 0.018 12.06 PFO/DeCo 8.10 0.010 62.55 26.7 0.020


( )2

210

Pa

COP −×

( )2

4

-1 -2

10

mol s m

COJ ×

⋅ ⋅ ( )

2

-1CO AmAmol mol

maxα

⋅

( )2

210

Pa

COP −×

( )2

4

-1 -2

10

mol s m

COJ ×

⋅ ⋅ ( )

2

-1CO AmAmol mol

maxα

⋅

5.08 PFO/DeCo 5.01 0.003 20.57 17.4 0.011 5.22 PFO/DeCo 4.49 0.005 20.75 17.8 0.010 7.91 PFO/DeCo 7.26 0.008 30.88 22.2 0.016 7.94 PFO/DeCo 7.32 0.008 41.00 28.2 0.016 10.48 PFO/DeCo 9.47 0.007 41.11 26.2 0.017 10.51 PFO/DeCo 9.50 0.006 49.65 31.3 0.018 15.54 PFO/DeCo 13.9 0.008 67.11 33.7 0.034 15.55 PFO/DeCo 14.7 0.008 67.20 31.7 0.022

20.41 15.9 0.009 68.41 37.4 0.026 20.53 16.4 0.010 68.44 35.0 0.021

59

60


( )2

210

Pa

COP −×

( )2

4

-1 -2

10

mol s m

COJ ×

⋅ ⋅ ( )

2

-1CO AmAmol mol

maxα

⋅

( )2

210

Pa

COP −×

( )2

4

-1 -2

10

mol s m

COJ ×

⋅ ⋅ ( )

2

-1CO AmAmol mol

maxα

⋅

2.50 PFO/DeCo 3.06 0.003 11.61 PFO/DeCo 12.8 0.014 4.58 PFO/DeCo 5.35 0.006 16.95 16.5 0.021 6.60 PFO/DeCo 7.52 0.008 31.56 26.0 0.025 8.52 PFO/DeCo 10.0 0.011 56.87 36.7 0.022


( )2

210

Pa

COP −×

( )2

4

-1 -2

10

mol s m

COJ ×

⋅ ⋅ ( )

2

-1CO AmAmol mol

maxα

⋅

( )2

210

Pa

COP −×

( )2

4

-1 -2

10

mol s m

COJ ×

⋅ ⋅ ( )

2

-1CO AmAmol mol

maxα

⋅

5.42 PFO/DeCo 6.26 0.004 15.46 PFO/DeCo 17.4 0.012 7.06 PFO/DeCo 8.82 0.006 16.54 PFO/DeCo 19.0 0.013 10.37 PFO/DeCo 11.3 0.008 20.42 18.8 0.012 10.43 PFO/DeCo 11.0 0.007 20.48 18.9 0.011 12.27 PFO/DeCo 13.9 0.009 25.59 25.6 0.017 13.61 PFO/DeCo 13.8 0.010 35.30 32.3 0.018 13.86 PFO/DeCo 16.3 0.010 45.53 37.7 0.017 15.23 PFO/DeCo 17.2 0.011 55.70 42.1 0.017 15.42 PFO/DeCo 15.8 0.010

60

60

61


( )2

210

Pa

COP −×

( )2

4

-1 -2

10

mol s m

COJ ×

⋅ ⋅ ( )

2

-1CO AmAmol mol

maxα

⋅

( )2

210

Pa

COP −×

( )2

4

-1 -2

10

mol s m

COJ ×

⋅ ⋅ ( )

2

-1CO AmAmol mol

maxα

⋅

2.53 PFO/DeCo 3.95 0.002 15.64 PFO/DeCo 20.6 0.013 7.74 PFO/DeCo 10.5 0.006 17.59 22.5 0.015 9.15 PFO/DeCo 12.7 0.009 17.62 23.2 0.010 9.33 PFO/DeCo 12.8 0.008 22.64 25.2 0.013 10.62 PFO/DeCo 15.2 0.010 27.52 28.0 0.017 11.08 PFO/DeCo 15.2 0.010 32.98 32.7 0.017 12.68 PFO/DeCo 17.6 0.011 55.02 43.4 0.016 13.78 PFO/DeCo 18.1 0.011 67.09 48.8 0.015 15.26 PFO/DeCo 20.9 0.014

Table 2.A10 - Kinetic data of the reaction of 2CO with potassium glycinate at 0.999 M and 303 K.

( )2

210

Pa

COP −× ( )

2

4

-1 -2

10

mol s m

COJ ×

⋅ ⋅ ( )

2

-1CO AmAmol mol

maxα

⋅

( )2

210

Pa

COP −× ( )

2

4

-1 -2

10

mol s m

COJ ×

⋅ ⋅ ( )

2

-1CO AmAmol mol

maxα

⋅

3.42 PFO/DeCo 4.99 0.003 10.54 PFO/DeCo 15.5 0.010 5.23 PFO/DeCo 7.68 0.005 11.59 PFO/DeCo 16.9 0.011 6.94 PFO/DeCo 10.0 0.007 17.70 24.7 0.016 8.59 PFO/DeCo 11.9 0.007 27.91 34.7 0.016 9.23 PFO/DeCo 13.4 0.009 57.65 54.7 0.020

61

62


( )2

210

Pa

COP −×

( )2

4

-1 -2

10

mol s m

COJ ×

⋅ ⋅ ( )

2

-1CO AmAmol mol

maxα

⋅

( )2

210

Pa

COP −×

( )2

4

-1 -2

10

mol s m

COJ ×

⋅ ⋅ ( )

2

-1CO AmAmol mol

maxα

⋅

2.77 PFO/DeCo 3.35 0.001 15.16 PFO/DeCo 19.6 0.006 4.76 PFO/DeCo 6.11 0.002 15.51 PFO/DeCo 18.6 0.006 6.74 PFO/DeCo 8.33 0.003 16.07 PFO/DeCo 18.4 0.006 7.71 PFO/DeCo 9.33 0.003 16.47 PFO/DeCo 21.2 0.006 8.31 PFO/DeCo 9.23 0.004 21.75 20.4 0.007 8.61 PFO/DeCo 11.6 0.004 21.78 19.8 0.007 8.74 PFO/DeCo 9.67 0.003 24.81 24.2 0.008 10.49 PFO/DeCo 13.9 0.004 25.00 24.1 0.008 10.78 PFO/DeCo 12.4 0.004 26.84 27.4 0.009 10.81 PFO/DeCo 12.1 0.004 36.88 35.0 0.009 12.56 PFO/DeCo 15.7 0.005 47.01 40.2 0.007 13.07 PFO/DeCo 18.2 0.006 57.00 47.9 0.008 13.26 PFO/DeCo 13.9 0.005 77.29 55.9 0.010

62

63


( )2

210

Pa

COP −×

( )2

4

-1 -2

10

mol s m

COJ ×

⋅ ⋅ ( )

2

-1CO AmAmol mol

maxα

⋅

( )2

210

Pa

COP −×

( )2

4

-1 -2

10

mol s m

COJ ×

⋅ ⋅ ( )

2

-1CO AmAmol mol

maxα

⋅

3.34 PFO/DeCo 5.12 0.001 13.53 PFO/DeCo 21.9 0.006 5.20 PFO/DeCo 7.51 0.002 14.33 PFO/DeCo 21.5 0.007 5.28 PFO/DeCo 7.01 0.002 15.36 PFO/DeCo 22.9 0.007 8.37 PFO/DeCo 13.3 0.004 17.34 27.3 0.007 9.31 PFO/DeCo 15.5 0.005 17.50 29.4 0.008 10.06 PFO/DeCo 13.8 0.005 19.42 29.4 0.008 10.24 PFO/DeCo 14.4 0.003 19.47 28.0 0.009 11.09 PFO/DeCo 16.6 0.005 21.52 26.8 0.008 11.30 PFO/DeCo 15.8 0.005 24.52 30.3 0.007 11.32 PFO/DeCo 15.3 0.005 29.47 33.9 0.007 12.52 PFO/DeCo 17.1 0.005 49.46 52.5 0.008 12.54 PFO/DeCo 18.5 0.005 69.86 66.3 0.007 13.42 PFO/DeCo 19.6 0.006


( )2

210

Pa

COP −×

( )2

4

-1 -2

10

mol s m

COJ ×

⋅ ⋅ ( )

2

-1CO AmAmol mol

maxα

⋅

( )2

210

Pa

COP −×

( )2

4

-1 -2

10

mol s m

COJ ×

⋅ ⋅ ( )

2

-1CO AmAmol mol

maxα

⋅

5.44 PFO/DeCo 10.6 0.003 11.52 PFO/DeCo 21.9 0.007 7.44 PFO/DeCo 14.8 0.005 14.34 DeCo 23.8 0.008 9.51 PFO/DeCo 19.1 0.006

63

64


( )2

210

Pa

COP −×

( )2

4

-1 -2

10

mol s m

COJ ×

⋅ ⋅ ( )

2

-1CO AmAmol mol

maxα

⋅

( )2

210

Pa

COP −×

( )2

4

-1 -2

10

mol s m

COJ ×

⋅ ⋅ ( )

2

-1CO AmAmol mol

maxα

⋅

3.43 PFO/DeCo 5.45 0.001 21.61 35.6 0.007 6.37 PFO/DeCo 11.4 0.003 26.49 38.4 0.008 11.44 PFO/DeCo 19.0 0.004 31.32 44.1 0.006 16.46 PFO/DeCo 26.7 0.005 51.45 60.3 0.005

64

65

3. Carbon dioxide absorption kinetics in

potassium threonate 1

Abstract

The absorption of carbon dioxide in potassium threonate aqueous solutions is studied at

concentrations ranging from 0.1 to 3 M and temperatures from 293 to 313 K. This study

includes experimental density, viscosity, solubility of 2N O and absorption kinetics of

2CO (using a stirred cell reactor) data obtained for the various potassium threonate

solutions. The diffusion coefficients of 2CO and potassium threonate in the absorption

solutions are estimated using a modified Stokes-Einstein relation. 2N O solubility is

interpreted using the Schumpe (1993) model and 2CO physical solubility estimated.

Physical absorption experiments were performed in the stirred cell in order to determine

the physical mass transfer coefficients. The kinetics results are interpreted using both

the pseudo-first order and the DeCoursey approaches. It was concluded that 2CO

absorption in the aqueous potassium threonate solutions is well represented by

( )2 2


T

− − = ×

.

1 Portugal, A. F.; Magalhães, F. D.; Mendes, A., “Carbon dioxide absorption kinetics in potassium threonate”, Chem. Eng. Sci, 2008, 63(13), 3493-3503

66

3.1. Introduction

The constant emissions of CO2 (due to the burning of fossil fuels, coal and natural gas)

have originated the rise of atmospheric concentrations to such values that the earth’s

natural absorption processes have become inefficient (Hampe and Rudkevich, 2003).

For this reason, several technologies for the capture of 2CO are being continuously

developed, especially in the last years, as a consequence of the stringent environmental

regulations stated on the Kyoto protocol (Granite and O'Brien, 2005; Idem and

Tontiwachwuthikul, 2006; Yan et al., 2007). Among these, the absorption of 2CO in

alkanolamines is, at present, the most widely used technology in the chemical industry

(Holst et al., 2006; Idem and Tontiwachwuthikul, 2006). However, alkanolamines easily

degrade, especially in oxygenated environments, making necessary the development of

new absorbents for gas mixtures with significant oxygen concentrations, namely for flue

gas, life support systems and anesthetic gas circuits (Goff and Rochelle, 2006; Holst et

al., 2006; Hook, 1997; Mendes, 2000; Portugal et al., 2007; Supap et al., 2006).

Amino acids (or, more precisely, alkali salts of amino acids) have the same functional

group as alkanolamines (presenting therefore similar capacities and reaction rates with

2CO ) and are much more stable in the presence of oxygen (Holst et al., 2006; Kumar et

al., 2003c). Besides, due to the ionic nature of the solutions, they present lower

volatilities and higher surface tensions (Kumar et al., 2002). Nevertheless, precipitation

of the reaction products was observed during the absorption of 2CO in several

absorbent solutions based on amino acids (Hook, 1997; Kumar et al., 2003a).

Commercially, amino acids are being used mainly as promoters of the absorption of

carbon dioxide in carbonate-bicarbonate solutions (Jeffreys and Bull, 1964; Kohl and

Nielsen, 1997). Companies including Giammarco Vetrocoke, BASF, TNO and Exxon

use amino acids in their 2CO removal absorption liquids (Feron and Jansen, 2002;

Kumar et al., 2003c). Although the use of amino acids is becoming an attractive option,

data about the absorption of 2CO in their solutions is still scarce. The present work aims

to contribute for the characterization and understanding of amino acid salt solutions

(namely, aqueous solutions of potassium threonate) as 2CO absorbents.

67

The general ability of an amine based compound to absorb 2CO is related to its

molecular structure (Caplow, 1968; Hook, 1997; Penny and Ritter, 1983; Sartori and

Savage, 1983; Singh et al., 2007). Potassium threonate, shown in Figure 3.1, is a

relatively small amino acid salt, which makes it expectable to present better absorption

equilibrium and kinetics than bigger molecules (Singh et al., 2007); it is a primary

amine containing both potassium carboxylate and alcohol groups in its structure. It is

sterically hindered, but the α carbon (carbon adjacent to the amine group) is only

mono-substituted. It is expected to have the properties of the sterically hindered amines

(high capacity and relatively easy regeneration) (Hook, 1997; Sartori and Savage, 1983)

and simultaneously, since it is a primary amine and the α carbon is only mono-

substituted, it is likely to absorb 2CO at a reasonable rate, without precipitation at low

loadings (Hook, 1997). It is also expected to combine the properties conferred by the

potassium carboxylate group - good oxidation stability, low volatility and high surface

tension – with the good regeneration properties conferred by the alcohol group (Hook,

1997). For these reasons along with the fact of not being harmful for the health, it was

selected for being characterized for carbon dioxide absorption.

Figure 3.1 - Chemical structure of potassium threonate.

3.2. Reaction Mechanism

Generally it is accepted that the absorption of 2CO in amine based compounds with a

primary or secondary amine group occurs according to the so-called zwitterion

mechanism (Caplow, 1968; Derks et al., 2006; Kumar et al., 2003c; Portugal et al.,

2007), being the overall reaction rate, 2COr− , given by:

68

2 2

2

11i i

CO S CO

B Bi

kr C C

k

k C−

− =+∑

(1)

where 2k , 1k− and iBk are the zwitterion mechanism kinetic constants, SC is the

concentration of the amino acid salt, 2COC is the concentration of carbon dioxide in the

liquid and iBC are the concentrations of the bases that can deprotonate the zwitterion -

2H O , OH − and the amine itself ( R1R

2NH ) (Blauwhoff et al., 1984). However, since

potassium threonate is a weak base - p 9.100AK = at 25 ºC (Perrin, 1965) - the

contribution of OH − to the deprotonation of the zwitterion can be neglected (Kumar et

al., 2003c) as well as the parallel reaction of 2CO with OH − to form bicarbonate.

Additionally, in primary amines such as monoethanolamine (MEA) and potassium

glycinate usually the deprotonation of the zwitterion is relatively fast when compared to

the rate of the reverse reaction (Derks et al., 2006; Kumar et al., 2003c; Portugal et al.,

2007) and therefore, equation (1) is simplified to a second order reaction kinetics:

2 22CO S COr k C C− = (2)

To be thermodynamically consistent, the 2CO absorption rate should be expressed in

terms of activities rather than concentrations (Haubrock et al., 2007). However, for the

purposes of the present work – to describe the absorption in a single absorption system

– it is enough to account for the solution non-idealities by means of the semi-empirical

equation (8) (Cullinane and Rochelle, 2006):

( )expeffk k bI= (3)

where effk is the effective kinetic constant, corrected by the ionic strength of the

solution, b is a constant and I is the ionic strength given by 21

2 i iI C z= ∑ , where iC

and iz are respectively the molar concentration and the charge of ion i in solution.

69

3.3. Mass Transfer

The absorption of CO2 into lean potassium threonate solutions can be described by the

following equation (Danckwerts, 1970):

2

2

2

COCO L

CO

PN E k A

H= (4)

where 2CON is the molar flow of CO2 crossing the liquid surface with interfacial area

A , Lk is the physical mass transfer coefficient, 2COP is the CO2 partial pressure and

2COH is the Henry constant of CO2 in solution. The enhancement factor, E , is the ratio

between the amount of gas absorbed in the reactive liquid and the amount that would be

absorbed if no reaction took place. The enhancement factor is a function of the Hatta

number, Ha , and the infinite enhancement factor, E∞ (Danckwerts, 1970; Derks et al.,

2006). The dimensionless Hatta number is defined as:

2ov CO

L

k DHa

k= (5)

where ovk is the overall reaction kinetic constant (2 2ov CO COk r C= − ) and

2COD is the

diffusion coefficient of CO2 in solution. According to the penetration theory (Higbie,

1935), the infinite enhancement factor can be estimated by the following equation:

2

2 2

2

310CO S S

COS COS

CO

D C DE

PD D

Hν

∞×= + (6)

where SD is the amino acid salt diffusion coefficient and Sν is the stoichiometric

coefficient. Factor 103 is there to convert -3mol dm⋅ (M) into -3mol m⋅ .

If Ha is sufficiently lower than E∞ , fast pseudo-first order (PFO) reaction regime can

be assumed (Danckwerts, 1970; Derks et al., 2006; Portugal et al., 2007):

3 0.1Ha E∞< <∼ (7)

In this case, diffusion and reaction occur in parallel in the liquid film. The enhancement

factor can be considered equal to the Hatta number and the gas absorption rate becomes,

therefore, independent of the physical mass transfer coefficient.

70

If E Ha∞ << , instantaneous reaction regime can be considered and E E∞= . In this

situation, the enhancement to the mass transfer is determined by the diffusion of the

reactants and do not depend on the reaction kinetic constant.

Between the limiting situations of fast pseudo-first order and instantaneous reaction

regime, there is the intermediate regime. According to DeCoursey, the enhancement

factor in the intermediate regime can be approximated as a function of the Hatta number

and the infinite enhancement factor (DeCoursey, 1974; Van Swaaij and Versteeg,

1992):

( ) ( )22 4

2 12 1 14 1

E HaHa HaE

E EE∞

∞ ∞∞

= − + + +− −−

(8)

3.4. Physical Properties

Since CO2 reacts with potassium threonate, its physical solubility and diffusivity in

solution need to be measured indirectly using a non-reactive gas with similar properties,

usually 2N O (Joosten and Danckwerts, 1972; Laddha et al., 1981).

Amino acid salt solutions are ionic in nature. For this reason, a “salting out” effect

needs to be taken into account when interpreting the solubility data of gases in these

solutions. At moderately high salt concentrations, this effect can be accounted for using

the Sechenov relation:

log Sw

HK C

H

= ⋅

(9)

where H and wH are respectively the Henry constants of the gas in the amino acid salt

solution and in water and K is the Sechenov constant, which can be calculated by

equation (12) (Schumpe, 1993; Weisenberger and Schumpe, 1996):

( )i G iK h h n= +∑ (10)

where ih and Gh are the ion and gas specific parameters and in is the valency number

of the ion.

It is generally accepted that the diffusion coefficient of a diffusant in solution can be

related to the solution viscosity, η , through modified Stokes-Einstein equation (Brilman

71

et al., 2001; Joosten and Danckwerts, 1972; Kumar et al., 2001; Versteeg and Van

Swaaij, 1988):

constantD αη = (11)

where α is a constant that depends on the pair diffusant/solvent.

It can be considered α = 0.8 to estimate the diffusion coefficient of 2N O in the

aqueous solutions of potassium threonate (Versteeg and Van Swaaij, 1988; Joosten and

Danckwerts, 1972; Brilman et al., 2001) and 0.6α = to estimate the diffusion

coefficient of the amino acid salt in solutions (Versteeg and Van Swaaij, 1988; Snijder

et al., 1993).

Gubbins et al. (1966) found that the ratio of the diffusivity of a gas in an electrolyte

solution to the diffusivity of the same gas in water does not vary significantly with the

nature of the diffusant. Therefore, it is reasonable to use the so-called 2N O analogy to

estimate the diffusion coefficient of 2CO in solutions.

2 2

2 2, ,

N O CO

N O w CO w

D D

D D= (12)

To estimate the diffusion coefficient of the salt at infinite dilution, 0SD , the Nernst

equation for the diffusion in electrolyte solutions can be applied (Poling et al., 2001):

( ) ( )( ) ( )

0

2 0 0

1 1

1 1S

RT z zD

F λ λ+ −

+ −

+ = +

(13)

where F is the Faraday constant, z+ and z− are the valencies of the cation and anion

respectively and 0λ+ and 0λ− are the ionic conductances of the cation and anion

respectively at infinite dilution.

The physical mass transfer coefficient, Lk , is related to the pair gas/solution and to the

apparatus where the mass transfer takes place through the empirical expression referred

by Versteeg et al. (1987):

3 42Sh Re Scc cc= (14)

72

where Sh, Re and Sc are respectively the Sherwood, Reynolds and Schmidt

dimensionless numbers and the constants 2c , 3c and 4c depend on the specific

apparatus. Performing experiments with a known and non-reactive gas/fluid pair (for

example 2 waterCO / or 2 solutionN O/ ) at different temperatures and stirring speeds, it is

possible to determine the constants of equation (14) and consequently, to estimate the

Lk of a given reactive system.

3.5. Experimental

Chemicals

The potassium threonate aqueous solutions were prepared by adding to the amino acid

an equimolar amount of potassium hydroxide ( )KOH in a volumetric flask filled up

with distilled and deionised water. It is important to notice that before the addition of

KOH , the amino acid exists in solution as a zwitterion (with the amine group

protonated). The addition of potassium hydroxide to form the potassium carboxylate

group will deprotonate the amine group enabling it to react with carbon dioxide (Kumar

et al., 2003c).

Density and Viscosity

Densities of potassium threonate solutions at 293, 298, 303 and 313 K were determined

using hydrometers series M100, ranges 1.000 to 1.100 and 1.100 to 1.200 ± 0.002

-1g ml⋅ .

Viscosities of the solutions of potassium threonate were determined using a standard

Cannon-Fenske viscosimeter.

73

N2O solubility

The procedure adopted to measure the solubility of 2N O in the amino acid salt

solutions is described in detail by Derks et al. (2005) and Portugal et al. (2007). The set-

up used is composed of two vessels with calibrated volumes; one for storing the 2N O

and the other for the absorbent solution, which is magnetically stirred. A known volume

of degassed solution is transferred to the absorbent vessel and the solution vapour

pressure, vapourP , recorded (pressure sensor from Druck, PMP4000, 0-350 mbar,

accuracy: 0.08% FS). A certain amount of 2N O is allowed to enter the absorbent tank

from the gas vessel and the initial pressure, 0P , recorded. The stirrer is then switched on

and the solution equilibrium is allowed to be established (the final pressure, eqP , is

recorded as well as the temperature, 0T ). The temperature is then set to a different

value, T , and a new equilibrium state is obtained; this procedure is repeated for the

temperatures at which the solubility is to be determined. The solution is weighed at the

end of the experiment. The amount of absorbed gas is calculated applying the ideal gas

law and the Henry coefficient for 2N O , 2N OH , is computed from the equation:

( ) ( ) ( )( ) ( ) ( )2

0 0

0

eq vapour LN O

Gvapour eq vapour

P T P T RVH T

VP P T P T P T

T T

− = − − −

(15)

where GV and LV are respectively the volume of gas and liquid in the absorbent vessel

and R is the universal gas constant. The solution vapour pressure at each temperature is

estimated by the following relation:

( ) ( )2 2

purevapour H O H OP T x P T= (16)

where 2H Ox is the molar fraction of water in solution. The water vapour pressure as a

function of the absolute temperature, P

H2Opure T( ), is obtained from the Antoine equation

(Poling et al., 2001).

Kinetic measurements

The experiments were performed in a stirred cell reactor with a smooth gas-liquid

interface operating batchwise with respect to the liquid phase and semi-continuously

74

with respect to the gas phase. Although experiments were performed in a different set-

up (much smaller – liquid volume: 50 cm3, reactor diameter: 3.87 cm, stirrer diameter: 2

cm), the followed procedure is the same of the one described by Derks et al. (2006) and

Portugal et al. (2007) and will be only briefly summarized here. Before starting the

experiment, the vapour-liquid equilibrium of the fresh solution of potassium threonate,

previously degassed, is established in the absorbent vessel and the vapour pressure,

vapourP , recorded (pressure sensor from Druck, PMP4000, 0-350 mbar, accuracy: 0.08%

FS). During the experiment, the pressure inside the stirred reactor is kept constant, SCP ,

using a pressure controller (Bronkhorst, P602-C, 0-200 mbar, accuracy: 0.5% FS) while

2CO from the gas vessel (filled with pure 2CO ) is being fed to it. All 2CO that is being

absorbed in the stirred cell comes from the gas vessel and therefore the flow of absorbed

2CO can be computed following the pressure decrease inside the gas vessel, GVdP

dt,

(pressure sensor from Druck, PMP4000, 0-2 bar, accuracy: 0.08% FS). Since the 2CO

absorption inside the stirred cell reactor is described by equation (9), after replacing

NCO2 by the pressure derivative and noticing that

P

CO2= P

SC− P

vapour, one obtains the

following expression:

2

SC vapourGV GVL

CO

P PV dPE k A

RT dt H

−= (17)

where GVV is the volume of the gas vessel. Finally, the experimental kinetic constant

can be extracted from the computed enhancement factor, depending on the absorption

reaction regime.

Figure 3.2 shows a sketch of the experimental setup.

75

Figure 3.2 – Experimental set-up sketch.

Physical mass transfer coefficient

The physical mass transfer coefficient, necessary for equation (17), was obtained from

experimental data of 2CO absorption in water (at different temperatures) and 2N O in

water and in potassium threonate solutions; the experimental set-up shown in Figure 3.2

was used. At a given temperature, a known volume of degassed water or absorbent

solution is placed in the stirred reactor, the vapour-liquid equilibrium allowed to be

established and the vapour pressure, vapourP , recorded. Then, a certain amount of 2CO or

2N O is admitted in the reactor, while the stirrer is switched off, and the initial pressure,

0P , recorded. The stirrer is then switched on, at a given stirring speed, and the pressure

history inside the reactor recorded. The procedure is repeated for different stirring

speeds given that a smooth gas-liquid interface is ensured. Only physical absorption

takes place, since the gas/liquid pairs are non-reactive (pairs 2CO /water and

2N O /absorbent aqueous solutions); for this reason, equation (9) becomes:

2 2

2 2

2

CO COGCO L CO

CO

dP PVN k A C

RT dt H

= = −

(18)

where 2COC is the absorbed gas concentration and VG is the volume of gas above the

liquid in the stirred reactor. Performing a mass balance to the stirred reactor one obtains:

( )2 2

2

0,CO CO GCO

L

P P VC

RT V

−= where

20, 0CO vapourP P P= − . Solving equation (18), it becomes:

76

2

2 2

2

2

0, 1ln

COL G G

CO CO

LCO G L

LCO

PRTV V V

H P RTk A t

RT H V VVH

+ − = − +

(19)

Note that equation (19) derived for the2CO physical absorption is also valid for 2N O .

3.6. Results and Discussion

Density and viscosity

Densities and viscosities of potassium threonate solutions at temperatures from 273 to

313 K and concentrations from 0.1 to 3.0 M were determined and are presented in Table

3.1.

N2O and CO2 solubility

The experimental solubility of 2N O and 2CO in water was experimentally determined

and it is given in Table 3.2. Although the Henry coefficients values obtained for 2N O

are in line with the ones reported in literature (Abu-Arabi et al., 2001), the values for

carbon dioxide in water are slightly below to what is reported by Abu-Arabi et al.

(2001).

The experimental solubility of 2N O in potassium threonate solutions is given in Table

3.2.

The results were interpreted using equation (9) and are presented graphically in Figure

3.3.

77

Table 3.1 – Densities and viscosities of potassium threonate solutions.

293 298 303 313 ( )KT

( )MSC ( )3kg m

ρ−⋅ ( )

3

1 1

10

kg m s

η− −

×

⋅ ⋅ ( )3kg m

ρ−⋅ ( )

3

1 1

10

kg m s

η− −

×

⋅ ⋅ ( )3kg m

ρ−⋅ ( )

3

1 1

10

kg m s

η− −

×

⋅ ⋅ ( )3kg m

ρ−⋅ ( )

3

1 1

10

kg m s

η− −

×

⋅ ⋅

0.1 1006 1.036 1004 0.929 1003 0.830 1000 0.683 0.3 1020 1.126 1018 1.007 1017 0.899 1014 0.785 0.6 1040 1.254 1038 1.146 1037 1.024 1034 0.888 1.0 1067 1.541 1065 1.355 1064 1.216 1061 1.080 2.0 1132 2.555 1131 2.276 1129 1.991 1125 1.577 3.0 1192 5.116 1190 3.984 1188 3.409 1184 2.827

Table 3.2 - Henry constants of 2N O and 2CO in water and in potassium threonate solutions. All values are experimental except for 2CO in

potassium threonate solutions that were computed based on Sechenov’s model - ( )3 1Pa m molH −⋅ ⋅ .

293 298 303 313 ( )KT

( )MSC 2N O 2CO 2N O 2CO 2N O 2CO 2N O 2CO

Water 3357 2442 3831 2771 4353 3132 5551 3953 0.1 3490 2520 3980 2854 4519 3219 5755 4048 0.3 3735 2683 4235 3026 4781 3400 6025 4244 0.6 4201 2949 4748 3305 5345 3691 6697 4557 1.0 4844 3344 5418 3717 6037 4119 7419 5009 2.0 6842 4578 7499 4987 8194 5418 9700 6349 3.0 9782 6268 10326 6690 10880 7126 12019 8046

77

78

Figure 3.3 – Sechenov plots of the 2N O solubility in potassium threonate solutions.

The specific parameters of Schumpe model for the cation and the gas (respectively, K

h +

and 2N Oh ) were taken from the work by Weisenberger and Schumpe (1996). These,

along with the Sechenov constants, enable to calculate the anion specific parameter,

Thh − , according to equation (10). Weisenberger and Schumpe (1996) also report the

2CO specific parameter, h

CO2, which allows to calculate the Sechenov constants and

consequently the Henry coefficients of 2CO in potassium threonate solutions. The

computed values of the Sechenov constants and the specific parameters of Schumpe

model are presented in Table 3.3. Figure 3.4 shows the computed anion specific

parameter as a function of the temperature.

79

Table 3.3 – Sechenov constants and specific parameters of Schumpe model for the

solubility of 2N O and 2CO in potassium threonate solutions.

T (K)

2N OK

( )3 -1dm mol⋅ 2N Oh *

( )3 -1dm mol⋅K

h + *

( )3 -1dm mol⋅ Th

h −

( )3 -1dm mol⋅2COh *

( )3 -1dm mol⋅

2COK

( )3 -1dm mol⋅

293 0.155 -0.0061 0.0753 -0.0155 0.136 298 0.145 -0.0085 0.0698 -0.0172 0.128 303 0.135 -0.0109 0.0646 -0.0189 0.119 313 0.116 -0.0157

0.0922

0.0552 -0.0223 0.103 * - Values taken from Weisenberger and Schumpe (1996)

Figure 3.4 – Threonate anion specific parameter as a function of temperature.

It was expected the ion specific parameters to be constant with temperature

(Weisenberger and Schumpe, 1996), however Th

h − clearly decreases with temperature

( )40.368 9.98 10Th

h T−−= − × , as it is shown in Table 3.3 and Figure 3.4. For this reason,

instead of taking the mean value of Th

h − (0.0662 3 -1dm mol⋅ ) to calculate the 2CO

solubility, 2COK was computed for each temperature by the following expression:

2 2 2 22 2CO N O N O COK K h h= − + ; results are presented in Table 3.3. The computed Henry

coefficients of 2CO in potassium threonate solutions are shown in Table 3.2.

80

Gas and ion diffusion coefficients

To estimate the diffusion coefficient of 2N O and 2CO in potassium threonate solutions,

equations (14) (with α = 0.8 ) and (12) were applied. Results are presented in Table 3.4.

The diffusion coefficients of potassium threonate in potassium threonate solutions were

estimated using the Stokes-Einstein relation – equation (14) – with 0.6α = and the

Nernst equation – equation (13). The ionic conductance at infinite dilution of the cation

K + , 0λ+ , as a function of temperature was computed based on the work by Fell and

Hutchiso (1971). The ionic conductance of the threonate anion at 298 K was linearly

interpolated using values of 0λ− available in literature for similar anions with molar

masses respectively lower and higher than the threonate anion: glycinate and aspartate

(Miyamoto and Schmidt, 1933). The temperature dependence was assumed to be linear

and equal to the one of aspartate, being 0λ− at 273 K of aspartate obtained from the work

by Hoskins et al. (1930). Table 3.4 presents the computed diffusion coefficients of

potassium threonate in the potassium threonate solutions.

Table 3.4 - Diffusion coefficient of 2N O , 2CO and potassium threonate in potassium

threonate solutions computed based on the Stokes-Einstein relation - ( )10 2 110 m sD −× ⋅

( )MSC 2N O 2CO Potassium

threonate 2N O 2CO Potassium threonate

293 K 298 K 0.1 15.2 16.6 9.47 17.3 18.6 10.6 0.3 14.2 15.6 9.01 16.2 17.5 10.1 0.6 13.0 14.3 8.44 14.6 15.7 9.35 1.0 11.0 12.1 7.46 12.8 13.8 8.45 2.0 7.37 8.07 5.51 8.42 9.09 6.19 3.0 4.23 4.63 3.63 5.38 5.81 4.43

303 K 313 K 0.1 19.7 21.0 11.8 25.2 26.1 14.4 0.3 18.5 19.7 11.3 22.5 23.4 13.2 0.6 16.7 17.7 10.4 20.4 21.2 12.3 1.0 14.5 15.4 9.42 17.4 18.1 10.9 2.0 9.79 10.4 7.01 12.9 13.4 8.72 3.0 6.36 6.77 5.08 8.08 8.38 6.14

81

Physical mass transfer coefficient

The physical mass transfer coefficient of 2CO in water was determined for the studied

temperatures and for stirring speeds ranging from 75 to 200 rpm. Physical mass transfer

coefficients of 2N O in water and in solutions of 1 M and 3 M were measured at

N = 200 rpm and 298 K. Equation (14) was fitted to the experimental data and it was

verified that the constants are within the usual values for stirred cell reactors (Versteeg

et al., 1987):

2 0.778 0.390Sh 6.33 10 Re Sc−= × (20)

The experimental and predicted Lk values differ less than 7 %. Since the physical

properties of the solutions are known, it is possible to extrapolate the value of Lk for the

potassium threonate solutions using equation (20). These values are presented in Table

3.5.

Table 3.5 - Physical mass transfer coefficient of 2CO in potassium threonate solutions,

computed based on equation (20) - ( )6 -110 m sLk × ⋅

( )KT

( )MSC 293 298 303 313

0.1 17.1 19.1 21.4 26.4 0.3 16.0 17.9 20.1 23.5 0.6 14.6 16.1 18.0 21.2 1.0 12.3 14.0 15.7 18.0 2.0 8.09 9.10 10.4 13.2 3.0 4.49 5.68 6.62 8.10

Kinetic Measurements

The experimental carbon dioxide flux data, 2COJ , as a function of the carbon dioxide

partial pressure, 2COP , for all temperatures and potassium threonate concentrations

studied are presented in appendix.

82

The carbon dioxide absorption performance of potassium threonate at 1 M and 298 K

was compared to the absorption performance of a primary and a secondary amines –

potassium glycinate and diethanolamine (DEA), respectively – obtained using the same

set-up and method. Results, shown in Figure 3.5, confirm that although potassium

threonate absorbs slower than potassium glycinate, it is faster than DEA. Looking to the

p AK values, this was an expected result since ,p 8.883A DEAK = < ,

p 9.100A Th

K − = <

,p 9.7775

A GlyK − = and since DEA has a secondary amine group while potassium

glycinate has a non-sterically hindered primary amine group. This result also confirms

that potassium threonate is able to absorb 2CO at a considerable rate.

Figure 3.5 – Comparison of 2CO absorption flux in potassium threonate, potassium

glycinate and diethanolamine (DEA) solutions at 1 M and 298 K (all measurements

were performed in the setup presented in Figure 3.2).

The results presented in the appendix were treated using both the pseudo-first order

assumption and equation (8) (DeCoursey, 1974) to obtain the overall kinetic constant.

For the pseudo-first order approach, only experiments that obey condition (11)

( 10E Ha∞ > ) were taken into account, while for the DeCoursey approach only

83

experiments at carbon dioxide partial pressures lower than 20 mbar were used. Such

low partial pressure range was chosen because this is when, within the DeCoursey

model, the overall kinetic constant is obtained with higher precision, mainly due to the

uncertainty of the 2CO and potassium threonate diffusivity coefficients. Results of both

approaches are presented in Table 3.6. The overall kinetic constants computed using

both methods are in agreement within a 20% difference (which corresponds to a

maximum deviation of 10% in the Ha values). This difference is very acceptable taking

into consideration that experiments at low partial pressures can be strongly affected by

experimental errors: 2CO partial pressures vary from values close to the solution vapour

pressure (in the range of tens of mbar) to values in the same order of magnitude as the

pressure equipment accuracy (in the range of tenths of mbar). For this reason, it was

decided to use the results obtained by the DeCoursey approach for further analysis,

since they were obtained using more experimental values, therefore reducing the

associated experimental error.

Table 3.6 – Experimental overall kinetic constants using the PFO and the DeCoursey

(DC) approaches - ( )1sovk − .

( )KT

( )MSC 293 298 303 313

PFO DC PFO DC PFO DC PFO DC 0.1 305 251 238 246 254 306 --- 0.3 773 824 986 1010 1240 1320 --- 0.6 2220 2380 2430 2620 4280 4420 --- 1.0 3920 4560 7610 7090 5790 7090 11 400 11 100 2.0 22 600 23 200 27 300 27 800 39 800 45 500 --- 3.0 --- 139 000 120 000 --- ---

It can be considered that at low loadings, the only ions present in solution are potassium

cation and threonate anion, both monovalent (2 1z = ). Hence, SI C= and combining

equations (7) and (8), it can be written:

( )2 2 2 expov CO CO S Sk r C k C bC= − = (21)

Assuming that the kinetic constant follows the Arrhenius law, it is possible to make an

overall fit for all the temperatures and concentrations:

84

( )2,0 exp expov S S

Ak k C bC

T =

(22)

The resulting fit, Equation (23), was obtained by minimizing the sum of the relative

residues and it is shown in Figures 3.6 and 3.7 along with the experimental results.

( )8 35804.13 10 exp exp 0.90ov Sk C I

T

− = ×

(23)

Penny and Ritter (1983) and Versteeg et al. (1996) suggested that the rate of 2CO

absorption in amines is related to the amine p AK . The kinetic constants, k2, were

compared with the Brønsted plots drew by these authors; p AK values of threonate as a

function of temperature were extracted from Perrin (1965). They were found to be

considerably lower than the expected and the discrepancy increases with temperature.

One possible explanation for this is the molecular configuration. Although the amine

group is not connected to a tertiary carbon, potassium threonate is sterically hindered

due to the hydroxyethylene group connected to the α carbon. This may confer

instability to the carbamate formed, hence decreasing the absorption rate. AMP, which

is a primary sterically hindered amine (with the α carbon dimethylated), diverges even

more from the referred Brønsted plots – ,p 9.72A AMPK = (Perrin, 1965),

k

2,AMP= 555 dm3 ⋅ mol-1 ⋅s-1 (Saha et al., 1995), both at 298 K – supporting this

hypothesis.

Usually, the effect of the ionic strength on the reaction kinetic constant is much lower

than 0.9b = (Cullinane and Rochelle, 2006; Portugal et al., 2007). To confirm the

obtained value for b, it was prepared a 1 M potassium threonate solution with the ionic

strength modified by adding NaCl up to 1 M (I = 2 M) and obtained the corresponding

overall kinetic constant. The obtained overall kinetic constant, at 298 K, was ovk = 14

900 s-1, as shown in Figure 3.7. This value is in agreement with the proposed model

given by equation (23). Nevertheless, it must be taken into account that the overall

kinetic constants were extracted based on computed values of the 2CO diffusion

coefficients - hence any uncertainty on these values affects significantly the final

results.

85

Figure 3.6 – Logarithmic plot of the overall absorption kinetic constant as a function of

the potassium threonate concentration - Experimental values and model curves.

Figure 3.7 – Semi-log plot of the apparent absorption kinetic constant, app ov Sk k C= , as

a function of the solution ionic strength - Experimental values and model curves.

86

3.7. Conclusions

Potassium threonate was characterized for carbon dioxide absorption. Densities and

viscosities of aqueous solutions with concentrations from 0.1 to 3 M at 293, 298, 303

and 313 K were determined. The solubility of 2N O in these solutions was measured at

the same temperatures. The results were treated using the Schumpe model (Schumpe,

1993). Threonate specific parameter was found to vary linearly with temperature -

40.368 9.98 10Th

h T−−= − × . Physical solubility of 2CO in solutions was then computed.

Diffusion coefficients of 2N O and 2CO were estimated using a modified Stokes-

Einstein relation and the 2N O analogy. Diffusion coefficient of potassium threonate at

infinite dilution was estimated using the Nernst equation for the diffusion in electrolyte

solutions and the diffusion of the ions in solutions was estimated applying modified

Stokes-Einstein relation.

The physical mass transfer coefficient of 2N O and 2CO in water and of 2N O in

solutions was measured and its dependence on the system physical properties for the

used absorption reactor obtained ( )2 0.778 0.390Sh 6.33 10 Re Sc−= × .

Kinetic measurements were performed in a stirred cell reactor operating semi-

continuously. It was verified experimentally that the 2CO absorption rate in potassium

threonate solutions are within the rates found for other amines. Nevertheless, the

computed kinetic constants do not follow the Brønsted plot proposed by Penny and

Ritter (1983) and by Versteeg et al. (1996). This has to do with the molecular

configuration, which is likely to originate unstable carbamates, thus favoring the

equilibrium and regeneration but penalizing the kinetics.

Since the absorption overall kinetics depends directly on the 2CO diffusivity, a more

accurate determination of this parameter in potassium threonate solutions will improve

the accuracy of the absorption overall kinetics. Further studies of potassium threonate as

a carbon dioxide absorbent should consider the absorption equilibrium and regeneration.

87

3.8. Nomenclature

A Gas-liquid interfacial area, m2

iBC Concentrations of the bases that can deprotonate the zwitterion, M

2COC Absorbed gas concentration, -3mol m⋅

SC Amino acid salt concentration, M


Sd Stirrer diameter, m



F Faraday constant, 96500 -1C mol⋅


h Ion and gas specific constants in the Shumpe equation, 3 -1m mol⋅

H Henry coefficient, -3Pa mol m⋅ ⋅


2COJ Carbon dioxide absorption flux, -2 -1mol m s⋅ ⋅

K Sechenov constant, 3 -1dm mol⋅

1k− Zwitterion kinetic constant of the reverse reaction, s-1

2k Zwitterion kinetic constant of the reaction, M-1 ⋅s-1

appk Apparent rate constant defined as: app ov Sk k C= , M-1 ⋅s-1

iBk Zwitterion mechanism deprotonation rate constant by base, M-1 ⋅s-1


ovk Overall kinetic constant, s-1

N Stirrer speed, rps

2CON Carbon dioxide absorption flow, -1mol s⋅

in Valency number of the ions


PFO Pseudo-first order reaction regime

88

2COr− Rate of reaction, -3 -1mol m s⋅ ⋅


Re Reynolds number, ( )2

Re Sd Nρη

= , dimensionless

Sh Sherwood number, 2

Sh L S

CO

k d

D= , dimensionless

Sc Schmidt number, 2

ScCOD

ηρ

= , dimensionless

T Temperature, K

V Volume, m3

x Molar fraction, -1mol mol⋅

,z z+ − Valencies of the cation and anion

Greek symbols

α Constant from the modified Stokes-Einstein equation

máxα Maximum loading achieved in one experiment, mol

CO2⋅ mol

S-1

Sν Stoichiometric coefficient


ρ Solution density, -3kg m⋅ 0λ+ ,

0λ− Ionic conductances of the cation and anion at infinite dilution, cm2 ⋅ Ω−1

Subscripts

0 Initial

2CO Carbon dioxide

DEA Diethanolamine

eff Effective (after correcting for the ionic strength)

eq Equilibrium

final Final

G Gas phase

GV Gas vessel

K + Potassium cation

89

L Liquid phase

2N O Nitrous oxide

S Amino acid salt

SC Stirred cell

Th− Threonate anion

w Water

3.9. References

Abu-Arabi, M. K., Al-Jarrah, A. M., et al. (2001). "Physical Solubility and Diffusivity

of CO2 in aqueous Diethanolamine Solutions". Journal of Chemical and Engineering

Data, 46(3), 516-521.

Blauwhoff, P. M. M., Versteeg, G. F., et al. (1984). "A Study on the Reaction between

CO2 and Alkanolamines in Aqueous-Solutions". Chemical Engineering Science, 39(2),

207-225.

Brilman, D. W. F., van Swaaij, W. P. M., et al. (2001). "Diffusion coefficient and

solubility of isobutene and trans-2-butene in aqueous sulfuric acid solutions". Journal of


Caplow, M. (1968). "Kinetics of carbamate formation and breakdown". Journal of the


Cullinane, J. T. and Rochelle, G. T. (2006). "Kinetics of carbon dioxide absorption into

aqueous potassium carbonate and piperazine". Industrial & Engineering Chemistry

Research, 45(8), 2531-2545.

Danckwerts, P. (1970). "Gas-Liquid Reactions". McGraw-Hill Book Company.

Decoursey, W. J. (1974). "Absorption with Chemical-Reaction - Development of a New

Relation for Danckwerts Model". Chemical Engineering Science, 29(9), 1867-1872.

Derks, P. W., Hogendoorn, K. J., et al. (2005). "Solubility of N2O in and density,

viscosity, and surface tension of aqueous piperazine solutions". Journal of Chemical and


90

Derks, P. W. J., Kleingeld, T., et al. (2006). "Kinetics of absorption of carbon dioxide in

aqueous piperazine solutions". Chemical Engineering Science, 61(20), 6837-6854.

Fell, C. J. D. and Hutchiso, H. P. (1971). "Diffusion Coefficients for Sodium and

Potassium Chlorides in Water at Elevated Temperatures". Journal of Chemical and

Engineering Data, 16(4): 427-429.


and dedicated absorption liquids: performances and prospects". Separation and



catalyzed degradation of monoethanolamine in CO2 capture processes". Industrial &


Granite, E. J. and O'Brien, T. (2005). "Review of novel methods for carbon dioxide

separation from flue and fuel gases". Fuel Processing Technology, 86(14-15), 1423-

1434.

Gubbins, K. E., Bhatia, K. K., et al. (1966). "Diffusion of Gases in Electrolytic

Solutions". American Institute of Chemical Engineers Journal, 12(3), 548.

Hampe, E. M. and Rudkevich, D. M. (2003). "Exploring reversible reactions between

CO2 and amines". Tetrahedron, 59(48), 9619-9625.

Haubrock, J., Hogendoorn, J. A., et al. (2007). "The applicability of activities in kinetic

expressions: a more fundamental approach to represent the kinetics of the system CO2-

OH- in terms of activities". Chemical Engineering Science, 62(21), 5753-5769.

Higbie, R. (1935). "The rate of absorption of a pure gas into a still liquid during a short

time of exposure". Transactions of the American Institute of Chemical Engineers, 31,

365–389.

Holst, J., Politiek, P. P., et al. (2006). "CO2 capture from flue gas using amino acid salt

solutions". GHGT-8 proceedings, Norway.

91


carbon dioxide scrubbing compounds". Industrial & Engineering Chemistry Research,

36(5), 1779-1790.

Hoskins, W. M., Randall, M., et al. (1930). "The conductance and activity coefficients

of glutamic and aspartic acids and their monosodium salts". Journal of Biological

Chemistry, 88(1), 215-239.

Idem, R. and Tontiwachwuthikul, P. (2006). "Preface for the special issue on the

capture of carbon dioxide from industrial sources: Technological developments and

future opportunities". Industrial & Engineering Chemistry Research, 45(8), 2413-2413.

Jeffreys, G. V. and Bull, A. F. (1964). "The effect of glycine additive on the rate of

absorption of carbon dioxide in sodium glycinate solutions". Transactions of the

Institution of Chemical Engineers and the Chemical Engineer, 42, 118-125.

Jensen, A., Jensen, J. B., et al. (1952). "Studies on Carbamates .6. The Carbamate of

Glycine". Acta Chemica Scandinavica, 6(3), 395-397.

Joosten, G. E. H. and Danckwerts, P. V. (1972). "Solubility and Diffusivity of Nitrous-

Oxide in Equimolar Potassium Carbonate-Potassium Bicarbonate Solutions at 25 ºC and

1 Atm". Journal of Chemical and Engineering Data, 17(4), 452.

Kohl, A. L. and Nielsen, R. B. (1997). "Gas Purification". Houston: Gulf Publishing

Company.

Kumar, P., Hogendoorn, J., et al. (2001). "Density, viscosity, solubility, and diffusivity

of N2O in aqueous amino acid salt solutions". Journal of Chemical and Engineering

Data, 46(6), 1357-1361.

Kumar, P., Hogendoorn, J., et al. (2003a). "Equilibrium solubility of CO2 in aqueous

potassium taurate solutions: Part 1. Crystallization in carbon dioxide loaded aqueous

salt solutions of amino acids". Industrial & Engineering Chemistry Research, 42(12),

2832-2840.

92

Kumar, P., Hogendoorn, J., et al. (2003b). "Equilibrium solubility of CO2 in aqueous

potassium taurate solutions: Part 2. Experimental VLE data and model". Industrial &

Engineering Chemistry Research 42(12), 2841-2852.

Kumar, P., Hogendoorn, J., et al. (2003c). "Kinetics of the reaction of CO2 with aqueous

potassium salt of taurine and glycine". American Institute of Chemical Engineers

Journal, 49(1), 203-213.

Kumar, P. S., Hogendoorn, J. A., et al. (2002). "New absorption liquids for the removal

of CO2 from dilute gas streams using membrane contactors". Chemical Engineering

Science, 57(9), 1639-1651.

Laddha, S. S., Diaz, J. M., et al. (1981). "The N2O Analogy - the Solubilities of CO2 and

N2O in Aqueous-Solutions of Organic-Compounds". Chemical Engineering Science,

36(1), 228-229.



anaesthetic closed loop". Applied Cardiopulmonary Pathophysiology, 9(2), 156-163.

Miyamoto, S. and Schmidt, C. L. A. (1933). "Transference and conductivity studies on

solutions of certain proteins and amino acids with special reference to the formation of

complex ions between the alkaline earth elements and certain proteins". Journal of

Biological Chemistry, 99(2), 335-358.


Dioxide and Primary Amines". Journal of the Chemical Society-Faraday Transactions I,

79, 2103-2109.

Perrin, D. (1965). "Dissociation Constants of Organic Bases in Aqueous Solutions".

Butterworth, London.

Poling, B. E., Prausnitz, J. M., et al. (2001). "The Properties of Gases and Liquids".

McGraw-Hill International Editions.

93

Portugal, A. F., Derks, P. W. J., et al. (2007). "Characterization of potassium glycinate

for carbon dioxide absorption purposes". Chemical Engineering Science, 62(23), 6534-

6547.

Saha, A. K., Bandyopadhyay, S. S., et al. (1995). "Kinetics of Absorption of CO2 into

Aqueous-Solutions of 2-Amino-2-Methyl-1-Propanol". Chemical Engineering Science,

50(22), 3587-3598.


from Gases". Industrial & Engineering Chemistry Fundamentals, 22(2), 239-249.

Schumpe, A. (1993). "The Estimation of Gas Solubilities in Salt-Solutions". Chemical


Singh, P., Nierderer, J. P. M., et al. (2007). "Structure and activity relationships for

amine based CO2 absorbents". International Journal of Greenhouse Gas Control, 1(1), 5-

10.

Snijder, E. D., Teriele, M. J. M., et al. (1993). "Diffusion-Coefficients of Several

Aqueous Alkanolamine Solutions". Journal of Chemical and Engineering Data, 38(3),

475-480.

Supap, T., Idem, R., et al. (2006). "Analysis of monoethanolamine and its oxidative



combinations". Industrial & Engineering Chemistry Research, 45(8), 2437-2451.


Complex Reversible Chemical-Reactions in Gas-Liquid Systems - an Overview".


Versteeg, G. F., Blauwhoff, P. M. M., et al. (1987). "The Effect of Diffusivity on Gas-

Liquid Mass-Transfer in Stirred Vessels - Experiments at Atmospheric and Elevated

Pressures". Chemical Engineering Science, 42(5), 1103-1119.

94

Versteeg, G. F., Van Dijck, L. A. J., et al. (1996). "On the kinetics between CO2 and

alkanolamines both in aqueous and non-aqueous solutions. An overview". Chemical

Engineering Communications, 144, 113-158.


Gases (CO2, N2O) in Aqueous Alkanolamine Solutions". Journal of Chemical and


Weisenberger, S. and Schumpe, A. (1996). "Estimation of gas solubilities in salt

solutions at temperatures from 273 K to 363 K". American Institute of Chemical


Yan, S. P., Fang, M. X., et al. (2007). "Experimental study on the separation of CO2

from flue gas using hollow fiber membrane contactors without wetting". Fuel

Processing Technology, 88(5), 501-511.

3.A. Experimental kinetic data

The experimental 2CO flux in aqueous potassium threonate solutions, 2COJ , as a

function of the 2CO partial pressure for all the concentrations and temperatures studied

is presented in Tables 3.A1 to 3.A6 along with he maximum loading reached at the end

of each experiment, maxα . All experiments started with fresh solutions.

Table 3.A1 – Flux of 2CO in 3 M potassium threonate solutions as a function of the

2CO partial pressure, at 298 K.

( )2

210 PaCOP −× ( )2

4 -1 -210 mol s mCOJ × ⋅ ⋅ ( )2

-1CO Smol molmáxα ⋅

0.98 1.42 0.00067 2.62 3.41 0.0016 3.37 4.59 0.0021 7.97 8.02 0.0052 9.98 10.1 0.0067 18.8 16.0 0.0074 24.0 17.9 0.0089 48.8 26.3 0.013 73.0 29.1 0.014

95

Table 3.A2 – Flux of 2CO in 0.1 M potassium threonate solutions as a function of the 2CO partial pressure, at 293, 298 and 303 K.

293 K 298 K 303 K

( )2

210

Pa

COP −×

( )2

4

-1 -2

10

mol s m

COJ ×

⋅ ⋅ ( )

2

-1CO Smol mol

maxα

⋅

( )2

210

Pa

COP −×

( )2

4

-1 -2

10

mol s m

COJ ×

⋅ ⋅ ( )

2

-1CO Smol mol

maxα

⋅

( )2

210

Pa

COP −×

( )2

4

-1 -2

10

mol s m

COJ ×

⋅ ⋅ ( )

2

-1CO Smol mol

maxα

⋅

1.70 0.481 0.0059 1.57 0.312 0.0047 1.05 0.260 0.0036 3.52 0.921 0.012 2.06 0.522 0.0073 2.32 0.516 0.0073 6.22 1.62 0.021 4.14 0.973 0.017 3.84 1.03 0.021 10.7 1.76 0.041 6.97 1.52 0.020 7.48 1.64 0.026 12.7 2.02 0.044 12.5 2.29 0.029 20.0 3.26 0.038 15.8 2.31 0.042 17.0 2.79 0.043 32.0 4.20 0.048 26.3 2.95 0.035 42.4 4.17 0.051 57.5 5.06 0.049 51.2 4.15 0.045 68.5 4.86 0.064 106 5.63 0.091 75.8 4.01 0.054


293 K 298 K 303 K

( )2

210

Pa

COP −×

( )2

4

-1 -2

10

mol s m

COJ ×

⋅ ⋅ ( )

2

-1CO Smol mol

maxα

⋅

( )2

210

Pa

COP −×

( )2

4

-1 -2

10

mol s m

COJ ×

⋅ ⋅ ( )

2

-1CO Smol mol

maxα

⋅

( )2

210

Pa

COP −×

( )2

4

-1 -2

10

mol s m

COJ ×

⋅ ⋅ ( )

2

-1CO Smol mol

maxα

⋅

2.83 1.14 0.0051 2.43 1.13 0.0073 2.35 1.05 0.0078 3.73 1.54 0.0087 4.07 1.72 0.0094 4.64 2.15 0.015 5.57 2.31 0.011 7.10 2.86 0.018 7.69 3.38 0.017 15.6 5.11 0.025 12.5 4.58 0.023 20.3 6.80 0.046 25.6 6.94 0.031 17.8 6.01 0.032 32.5 9.69 0.048 50.5 9.35 0.043 42.0 9.52 0.044 57.1 12.3 0.060 75.5 10.3 0.048 67.3 11.8 0.064 95

96


293 K 298 K 303 K

( )2

210

Pa

COP −×

( )2

4

-1 -2

10

mol s m

COJ ×

⋅ ⋅ ( )

2

-1CO Smol mol

maxα

⋅

( )2

210

Pa

COP −×

( )2

4

-1 -2

10

mol s m

COJ ×

⋅ ⋅ ( )

2

-1CO Smol mol

maxα

⋅

( )2

210

Pa

COP −×

( )2

4

-1 -2

10

mol s m

COJ ×

⋅ ⋅ ( )

2

-1CO Smol mol

maxα

⋅

2.36 1.45 0.0037 3.68 2.09 0.0070 2.26 1.68 0.0075 3.27 2.06 0.0051 4.41 2.68 0.0088 4.33 3.23 0.012 3.91 2.42 0.0060 7.79 4.85 0.016 7.64 5.79 0.015 5.60 3.27 0.0081 12.9 6.95 0.023 19.7 10.8 0.038 15.5 7.70 0.019 17.8 9.03 0.028 32.9 15.7 0.042 25.6 10.9 0.027 43.4 16.4 0.042 57.5 20.3 0.052 75.3 18.1 0.045 68.2 20.3 0.048

Table 3.A5 – Flux of 2CO in 2 M potassium threonate solutions as a function of the 2CO partial pressure, at 293, 298 and 303 K.

293 K 298 K 303 K

( )2

210

Pa

COP −× ( )

2

4

-1 -2

10

mol s m

COJ ×

⋅ ⋅ ( )

2

-1CO Smol mol

maxα

⋅

( )2

210

Pa

COP −× ( )

2

4

-1 -2

10

mol s m

COJ ×

⋅ ⋅ ( )

2

-1CO Smol mol

maxα

⋅

( )2

210

Pa

COP −× ( )

2

4

-1 -2

10

mol s m

COJ ×

⋅ ⋅ ( )

2

-1CO Smol mol

maxα

⋅

3.12 2.82 0.0023 2.39 2.71 0.0030 2.49 3.18 0.0045 5.15 5.08 0.0043 5.21 5.06 0.0052 6.19 7.27 0.0057 7.45 6.81 0.0053 10.1 8.45 0.0087 11.5 13.5 0.0099 17.6 13.3 0.010 16.2 13.7 0.011 24.1 20.4 0.015 27.9 18.4 0.013 20.9 16.8 0.013 36.2 29.5 0.021 52.7 25.9 0.016 45.8 27.8 0.016 60.8 36.1 0.017 77.6 30.9 0.019 70.7 36.5 0.019

96

97

Table 3.A6 – Flux of 2CO in 1 M potassium threonate solutions as a function of the 2CO partial pressure, at 293, 298, 303 and 313 K.

293 K 298 K

( )2

210

Pa

COP −×

( )2

4

-1 -2

10

mol s m

COJ ×

⋅ ⋅ ( )

2

-1CO Smol mol

maxα

⋅

( )2

210

Pa

COP −×

( )2

4

-1 -2

10

mol s m

COJ ×

⋅ ⋅ ( )

2

-1CO Smol mol

maxα

⋅

2.55 1.88 0.0027 2.49 2.03 0.0047 4.16 2.93 0.0042 4.35 3.87 0.0083 5.60 3.90 0.0055 9.44 7.71 0.012 7.87 4.76 0.0070 14.6 10.8 0.016 11.6 7.38 0.011 19.3 11.9 0.017 16.7 9.87 0.015 44.8 22.2 0.032 27.0 13.9 0.020 69.7 26.5 0.038 52.0 20.9 0.032 119 32.3 0.038 76.7 23.6 0.035

303 K 313 K

( )2

210

Pa

COP −× ( )

2

4

-1 -2

10

mol s m

COJ ×

⋅ ⋅ ( )

2

-1CO Smol mol

maxα

⋅

( )2

210

Pa

COP −× ( )

2

4

-1 -2

10

mol s m

COJ ×

⋅ ⋅ ( )

2

-1CO Smol mol

maxα

⋅

2.02 1.80 0.0031 2.36 2.14 0.0024 2.86 2.77 0.0053 8.75 7.15 0.011 4.10 2.30 0.0066 32.7 22.0 0.032 5.17 3.90 0.0070 5.87 4.15 0.0066 10.4 8.12 0.016 22.5 16.2 0.023 36.8 20.6 0.031 60.7 27.2 0.041 110 35.6 0.036 97

Part III

101

4. Solubility of carbon dioxide in

aqueous solutions of amino acid salts 1

Abstract

The solubility of 2CO in aqueous solutions of potassium glycinate was measured in a

stirred reactor, at temperatures from 293 to 351 K, for amino acid salt concentrations

ranging between 0.1 and 3.0 M and 2CO partial pressures up to 46 10× Pa. 2CO

solubility in potassium threonate 1.0 M was also measured at 313 K. It was observed that

amino acid salt solutions can be very interesting for 2CO absorption purposes since they

present considerably high absorption capacities. Nevertheless, 2CO solubility in these

solutions does not change significantly for temperatures between 293 and 323 K, which

can be a draw back concerning the absorbent regeneration.

Potassium glycinate solubility data were interpreted using the thermodynamically sound

model proposed by Deshmukh and Mather (1981) and the empirical Kent and Eisenberg

(1976) model.

1 Portugal, A. F.; Sousa, J. L.; Magalhães, F. D.; Mendes, A., “Solubility of carbon dioxide in aqueous solutions of amino acid salts”, Chem. Eng. Sci., DOI: 10.1016/j.ces.2009.01.036

102

4.1. Introduction

The climate change due to the rising greenhouse gases concentrations in the atmosphere

became an unquestionable problem nowadays (Idem and Tontiwachwuthikul, 2006; IEA,

2008; UNFCCC, 2008). Since the Kyoto protocol, in December 1997, several stringent

environmental regulations are being proposed and recently (in January 2008) the

European Council (EC) stated as a key target to reduce 20 % greenhouse gases emissions

by 2020 (EC, 2008; Gibbins and Chalmers, 2008). Among the greenhouse gases, carbon

dioxide ( 2CO ) is the one released in larger extent by human activity (Idem and

Tontiwachwuthikul, 2006; UNFCCC, 2008), hence, much effort is being put on the

development of technologies for 2CO capture and storage. Three basic processes can be

applied for capturing 2CO from flue gases: oxyfuel combustion, pre-combustion, and

post-combustion (Idem and Tontiwachwuthikul, 2006; Metz et al., 2005). Although

apparently less efficient than pre- and oxyfuel combustion, post-combustion seems to be

the best solution to meet the exigent EC emission targets, because it can be easily

retrofitted to already existing equipment and power plants (Favre, 2007; Gibbins and

Chalmers, 2008).

Chemical absorption in liquid solutions is a proven technology for 2CO removal from a

variety of gas streams and it is considered the best available technology for post-

combustion flue gas treatment (Favre, 2007; Gibbins and Chalmers, 2008; Idem and

Tontiwachwuthikul, 2006). Nevertheless, the absorbent solutions commonly used

(alkanolamine solutions) undergo oxidative degradation, so they might not be suitable

for the specified separation due the high oxygen concentrations present in flue gas (Goff

and Rochelle, 2006; Hook, 1997; Supap et al., 2006). Because of their higher stability in

the presence of oxygen, growing interest is being given to amino acid salts solutions.

Amino acid salts are more resistant to oxidative degradation, have negligible volatilities

and their aqueous solution present viscosities and surface tensions similar to water (Holst

et al., 2006; Kumar et al., 2002; Kumar et al., 2003c; Portugal et al., 2007). Additionally,

they react with 2CO in the same way as alkanolamines, presenting therefore comparable

absorption kinetics and equilibrium capacities (Hook, 1997; Kumar et al., 2003b; Kumar

et al., 2003c; Portugal et al., 2007; Portugal et al., 2008; Song et al., 2006). Nevertheless,

103

precipitation of reaction products was observed during the absorption of 2CO in aqueous

solutions of amino acids salts (Hook, 1997; Kumar et al., 2003a).

Although amino acids are used for 2CO absorption in a number of industrial solutions

(Feron and Jansen, 2002; Kumar et al., 2003c), very few data are available in literature

for these systems. In the present work, the absorption capacity of potassium glycinate

towards 2CO is determined at different temperatures, using potassium glycinate aqueous

solutions with initial concentrations ranging from 0.1 to 3.0 -3mol dm⋅ and for 2CO

partial pressures up to 46 10× Pa. The absorption capacity of a solution of potassium

threonate, 1.0 -3mol dm⋅ , at 313 K was also measured for the same pressure range.

Several models are available in literature for representing the vapour – liquid equilibrium

in the 2CO – amines – water systems. They are basically divided into three types:

- Empirical models such as Kent and Eisenberg (1976) model, where the non-

idealities of the system are lumped in the equilibrium constants. Despite its

simplicity, the Kent-Eisenberg model has been successfully applied to a number

of 2CO absorption systems (Aroua and Salleh, 2004; Li and Shen, 1993; Park et

al., 2002; Tontiwachwuthikul et al., 1991) and has demonstrated to predict fairly

well the system behaviour for loadings between 0.2 and 0.7 -1mol mol2CO Amine⋅

(Kumar et al., 2003b; Weiland et al., 1993).

- Models based on the excess Gibbs energy, in which a term to account for the

electrostatic forces due to the presence of ions in solution is added to the

molecular Gibbs energy models. Examples of this approach are the Deshmukh

and Mather (1981) method, the electrolyte-NRTL model of Chen and Evans

(1986) and the models developed by Austgen et al. (1989) and Clegg and Pitzer

(1992).

- Models using an equation of state (EoS) to describe both liquid and vapour

phases. A term to account for the ionic interactions is also added to the molecular

EoS. Applications of the EoS models to 2CO absorption systems are quite recent.

104

Further details and applications can be found in the work of Fürst and Renon

(1993), Vallee et al. (1999), Li and Fürst (2000), Derks et al. (2005) and

Huttenhuis et al. (2008).

In the present work, two models are considered: the empirical Kent and Eisenberg (1976)

and the Deshmukh and Mather (1981). The model developed by Deshmukh and Mather

(1981) is thermodynamically sound and reasonably simple when compared to other

electrolyte-NRTL and EoS models. It considers the long-range electrostatic interactions

and short-range Van der Walls interactions between the chemical species to describe the

system non-idealities and is used with success to represent acid gases – amines – water

systems (Benamor and Aroua, 2005; Kumar et al., 2003b; Weiland et al., 1993;

HajiSulaiman et al., 1996; Liu et al., 1999; Ma'mun et al., 2006; Rascol et al., 1996;

Tobiesen et al., 2008).

4.2. Modelling

Amino acid salts containing a primary amine group (such as potassium glycinate and

potassium threonate) react with 2CO according to the following equilibrium reactions

(where 2R CH COO−≡ , for potassium glycinate):

Carbamate hydrolysis

2 2 3 RNHCOO H O RNH HCO− −+ + (1)

Amine deprotonation

3 2RNH RNH H+ ++ (2)

Bicarbonate formation

2 2 3CO H O HCO H− ++ + (3)

Carbonate formation

23 3HCO CO H− − ++ (4)

105

Water auto-ionization

2H O OH H− ++ (5)

The reaction kinetically dominant in the absorption system is the direct reaction between

2CO and the amino acid salt to form a carbamate and the protonated amino acid:

Direct reaction between 2CO and the amino acid salt

2 2 32RNH CO RNH RNHCOO+ −+ + (6)

However, reaction (6) can be written as a combination of the independent reactions (1),

(2) and (3) (Kumar et al., 2003b).

The equilibrium constants, K , of the above independent reactions can be expressed as

follows:

[ ]

2 32 3 RNH HCOcarb

w RNHCOO

RNH HCOK

aRNHCOO

γ γγ

−

−

−

−

=

(7)

[ ]

2

3

2

3

RNH HAmA

RNH

RNH HK

RNH

γ γγ

+

+

+

+

=

(8)

[ ]3

2

2

3

2

HCO HCO

w CO

HCO HK

CO a

γ γγ

− +− + = (9)

23

3

3

23

3

CO H

HCOHCO

CO HK

HCO

γ γγ

− +

−

−

− +

−

=

(10)

OH Hw

w

K OH Ha

γ γ− +− + = (11)

where [ ]i are the molar concentrations of species i in the liquid phase, iγ are the

respective activity coefficients and wa is the water activity. The equilibrium constant of

reaction (6) is given by (Kumar et al., 2003b):

2COov

AmA carb

KK

K K= (12)

106

Additionally to the equilibrium constants, the following mass conservation and charge

balance equations must be verified:

Amine mass balance

[ ] [ ]2 2 30RNH RNH RNHCOO RNH− + = + + (13)

CO2 mass balance

[ ] [ ] 22 2 3 30

RNH CO RNHCOO HCO COα − − − = + + + (14)

where α is the loading: moles of 2CO absorbed per mole of amino acid salt initially in

solution.

Charges balance

For amino acid salts, R is charged and, therefore, 2RNH is a mono-valent anion,

RNHCOO− is bivalent and 3RNH+ is neutral. The charge balance becomes then:

[ ] 22 3 32 2K H RNH RNHCOO HCO CO OH+ + − − − − + = + + + + (15)

Since the pressure range considered is always lower than 1× 105 Pa, the vapour phase can

be considered ideal (Smith et al., 1996) and the vapour-liquid equilibrium is described by

the Henry Law:

[ ]2 2 2CO COP H CO= (16)

where 2COP is the 2CO partial pressure and

2COH is the Henry coefficient of 2CO in

potassium glycinate solutions, obtained experimentally in a previous work (Portugal et

al., 2007).

To account for the liquid non-idealities, the Deshmukh-Mather method (Deshmukh and

Mather, 1981) can be applied. The Deshmukh-Mather method uses the following

equation, originally proposed by Guggenheim (1935), to calculate the activity

coefficients:

[ ]2

,

2.303ln 2

1i

i i jj solvent

Az Ij

Ba Iγ β

≠

−= ++ ∑ (17)

107

The first term of equation (17) accounts for the long-range interactions between species

and is based on the Debye-Hückel theory. The solution ionic strength, I , is defined as:

[ ] 21

2 jj

I j z= ∑ (18)

where jz is the ion charge and [ j ] is the molar concentration of species j. The Debye-

Hückel limiting slope, A , and the parameter B depend on the temperature, T , and on

the dielectric constant of the solvent, ε , as follows (Kumar et al., 2003b):

( ) 3 261.825 10A Tε −= × (19)

( ) 1 250.3B Tε −= (20)

The dielectric constant of the solvent (water) is given by ( )80 0.4 293Tε = − −

(Knowlton, 1941). Parameter a roughly corresponds to the effective size of the hydrated

ions (Weiland et al., 1993). The second term of equation (17) accounts for the short-

range interactions between molecular and ionic solutes by means of the adjustable binary

interaction parameters, ,i jβ .

In the present absorption system, the following 10 species are present in solution: 2H O ,

2CO , 3HCO− , 23CO − , 2RNH , K + , 3RNH+ , RNHCOO− , H + and OH − . Excluding the

solvent ( 2H O ), there are still 9 species which short range interactions are to be taken

into account. This leads to 36 binary interaction parameters, ,i jβ , resulting in an

intractable problem. To make it manageable, Weiland et al. (1993) proposed the

following assumptions that reduce the number of binary interaction parameters needed to

be fit:

- All interactions between like charged ions are neglected;

- All self-interactions of molecular species are neglected, except for the molecular amine

(in the present case, the protonated amine: 3RNH+ );

- All interactions with water and its self-ionization products (H + and OH − ) are

neglected;

- All interactions with 2CO and 23CO − are neglected.

The number of interaction parameters is consequently reduced to 8:

108

3 3RNH HCO−− , 3 3RNH RNH+ +− , 3K HCO+ −− , 2K RNH+ − , 3 2RNH RNH+ − ,

3K RNH+ +− , K RNHCOO+ −− and 3RNH RNHCOO+ −− .

Instead, Kent and Eisenberg (1976) considered the liquid phase non-idealities lumped in

the equilibrium constants. In their model, the equilibrium constants are only functions of

the species concentrations (iγ =1) and carbK and AmAK are used as fitting parameters.

Therefore, the equilibrium constants carbK and AmAK obtained by this method are

apparent constants.

4.3. Experimental

The aqueous solutions of the amino acid salts were prepared by adding to the amino acid

an equimolar amount of potassium hydroxide ( )KOH in a volumetric flask filled up

with distilled and deionized water. The concentrations of the solutions were checked by

titration with a standard solution of HCl 1 M.

The set-up used to determine the solubility of 2CO is shown in Figure 4.1. It is

composed of two vessels with calibrated volumes, one for storing the 2CO (237.04 mL)

and the other for the absorbent solution (110.56 mL), which is magnetically stirred. The

temperature, T , is controlled by means of an in-house developed thermostatic closet and

a thermostatic bath that controls the jacket temperature of the absorbent reactor.

Figure 4.1 - Experimental set-up sketch.

109

Using a setup with lower dimensions than usually found in the literature (Derks et al.,

2005) allows for the use of small quantities of the amino acid salt solutions, even though

it implies larger relative errors in the determination of the equilibrium loadings

(estimated to be lower than 10 %).

A known volume (about 50 mL) of fresh solution of amino acid salt, solV , previously

degassed is transferred to the absorbent vessel. The vapour-liquid equilibrium is

established and the vapour pressure, vapourP , recorded (pressure sensor from Druck,

PMP4000; accuracy: ± 0.28 210× Pa). On the meantime, the gas vessel is filled with

2CO and the pressure recorded, 2 , 0

kCOP (pressure sensor from Druck, PMP4000;

accuracy: ± 1.6 210× Pa); pressure inside the gas vessel is always lower than 3 510× Pa.

After that, a certain amount of 2CO is transferred from the gas vessel to the absorbent

vessel and the pressure inside the gas vessel is again recorded, 2 ,

kCO finP . Once the

equilibrium is attained (this happens when the pressure becomes constant inside the

absorbent vessel) the pressure of the absorbent vessel is recorded, keqP . More 2CO is then

admitted from the gas vessel into the absorbent vessel and a new equilibrium value is

achieved. It is assumed that the solution vapor pressure does not change with loading and

the amount of 2CO absorbed is computed from the ideal gas law:

2 2

2, 2,

, 0 ,1k k

CO CO fink kCO add CO add GV

P Pn n V

RT− −

= + (21)

( )2, 2,

keq vapourk k

CO abs CO add AV sol

P Pn n V V

RT

−= − − (22)

where GVV and AVV are respectively the volumes of the gas and absorbent vessels. For

amino acid salt concentrations higher than the ones used in the present work, the

assumption of constant vapor pressure might become unrealistic for loadings higher than

0.5 2

-1CO Smol mol⋅ , because of the change on the liquid composition.

The loading, kα , corresponding to each 2CO partial pressure ( )2 ,

k kCO eq eq vapourP P P= − is

then calculated:

[ ]2,

2 0

kCO abs sol

k

n V

RNHα = (23)

110

Sub and super scripts k denote the experimental stage.

The small size of the setup used (stirred reactor: 110.56 mL, liquid volume 50 mL≈ and

gas vessel: 237.04 mL) enables the characterization of the amino acid salt solutions using

small quantities of reactant within an error that should be smaller than 10 %.


Experimental method validation

Before and after measuring the solubility of 2CO in potassium glycinate, the set up was

tested with 2.5 M aqueous solution of monoethanolamine (MEA) and the results

compared with the ones reported in literature (Jones et al., 1959; Lee et al., 1974; Lee et

al., 1976; Shen and Li, 1992) – please see Figure 4.2. Experimental values of the

solubility of 2CO in MEA solutions are presented in Table 4.1.

Figure 4.2 – Semi-log plot of the solubility of 2CO in aqueous solutions of MEA 2.5 M,

at 313 K - comparison with results from literature.

111

Table 4.1 - Solubility of 2CO in aqueous solutions of MEA 2.5 M.

Run 1 - 2.51 MMEAC = Run 2 - 2.43 MMEAC =

( )2

210

Pa

COP −× ( )

2

-1CO Smol mol

α

⋅ ( )2

210

Pa

COP −× ( )

2

-1CO Smol molα ⋅

2.03 0.173 2.82 0.233 3.95 0.331 10.2 0.471 21.1 0.491 211 0.621 414 0.635 628 0.678

2CO solubility in potassium glycinate

The experimental values of the solubility of 2CO in the potassium glycinate solutions are

shown in Tables 4.2 to 4.4. The results for 2CO solubility in a 1.0 -3mol dm⋅ potassium

glycinate solution are presented in Figure 4.3 and Figure 4.4 shows the results for a 3.0

-3mol dm⋅ solution.

Figure 4.3 – Semi-log plot of the experimental solubility of 2CO in aqueous solutions of

potassium glycinate, 1.0 -3mol dm⋅ - comparison with the results from Song et al. (2006)

for an aqueous solution of sodium glycinate 1.06 -3mol dm⋅ , at 313 and 323 K.

112

Table 4.2 - Experimental solubility of 2CO in aqueous solutions of potassium glycinate 0.1 M.

293 K 303 K 313 K 323 K

( )2

210

Pa

COP −× ( )

2

-1CO Smol mol

α

⋅ ( )2

210

Pa

COP −× ( )

2

-1CO Smol mol

α

⋅ ( )2

210

Pa

COP −× ( )

2

-1CO Smol mol

α

⋅ ( )2

210

Pa

COP −× ( )

2

-1CO Smol mol

α

⋅

0.85 Run 1 0.332 0.89 Run 1 0.318 0.54 Run 1 0.213 6.20 Run 1 0.650 2.55 Run 1 0.607 4.60 Run 1 0.728 1.57 Run 1 0.428 83.2 Run 1 1.096 18.7 Run 1 0.867 90.88 Run 1 1.129 4.09 Run 1 0.620 388 Run 1 1.289 71.4 Run 1 0.958 4.02 Run 2 0.692 16.52 Run 1 0.864 3.10 Run 2 0.471 639 Run 1 1.209 6.93 Run 2 0.778 60.47 Run 1 1.044 32.2 Run 2 0.953 6.88 Run 2 0.740 12.9 Run 2 0.868 208.46 Run 1 1.179 361 Run 2 1.294 396 Run 2 1.071 23.3 Run 2 0.949 624.47 Run 1 1.320 5.30 Run 3 0.462 40.7 Run 3 1.002 572 Run 2 1.357 16.1 Run 3 0.794 448 Run 3 1.205 112 Run 3 1.152

310 Run 3 1.309

112

113

Table 4.3 – Experimental solubility of 2CO in aqueous solutions of potassium glycinate 1.0 M.

293 K 298 K 313 K 323 K 351 K

( )2

210

Pa

COP −× ( )

2

-1CO Smol mol

α

⋅ ( )2

210

Pa

COP −× ( )

2

-1CO Smol mol

α

⋅ ( )2

210

Pa

COP −× ( )

2

-1CO Smol mol

α

⋅ ( )2

210

Pa

COP −× ( )

2

-1CO Smol mol

α

⋅ ( )2

210

Pa

COP −× ( )

2

-1CO Smol mol

α

⋅

1.15 Run 1 0.358 0.95 Run 1 0.349 1.40 Run 1 0.302 10.3 Run 1 0.353 13.2 Run 1 0.137 9.54 Run 1 0.572 4.28 Run 1 0.538 17.0 Run 1 0.610 23.1 Run 1 0.524 14.0 Run 1 0.159 13.1 Run 1 0.590 23.2 Run 1 0.637 37.5 Run 1 0.665 158 Run 1 0.692 15.7 Run 1 0.200 21.5 Run 1 0.621 36.2 Run 1 0.666 79.6 Run 1 0.721 389 Run 1 0.776 18.7 Run 1 0.247 29.8 Run 1 0.643 616 Run 1 0.920 547 Run 1 0.899 3.45 Run 2 0.169 24.1 Run 1 0.303 38.8 Run 1 0.661 1.30 Run 2 0.360 5.60 Run 2 0.282 28.9 Run 1 0.375 55.5 Run 1 0.689 2.02 Run 2 0.400 8.90 Run 2 0.403 62.5 Run 1 0.452 211 Run 1 0.810 3.87 Run 2 0.479 30.1 Run 2 0.550 146 Run 1 0.539 617 Run 1 0.915 7.18 Run 2 0.535 108 Run 2 0.649 288 Run 1 0.604 0.37 Run 2 0.172 13.4 Run 2 0.582 437 Run 2 0.773 12.3 Run 2 0.129 1.34 Run 2 0.420 34.4 Run 2 0.645 13.0 Run 2 0.150 2.54 Run 2 0.490 305 Run 2 0.846 15.0 Run 2 0.200 9.62 Run 2 0.559 19.0 Run 2 0.268 34.6 Run 2 0.632 26.5 Run 2 0.349 174 Run 2 0.761 41.5 Run 2 0.441

253 Run 2 0.625

113

114

Table 4.4 – Experimental solubility of 2CO in aqueous solutions of potassium glycinate 3.0 M.

293 K 303 K 313 K 323 K

( )2

210

Pa

COP −× ( )

2

-1CO Smol mol

α

⋅ ( )2

210

Pa

COP −× ( )

2

-1CO Smol mol

α

⋅ ( )2

210

Pa

COP −× ( )

2

-1CO Smol mol

α

⋅ ( )2

210

Pa

COP −× ( )

2

-1CO Smol mol

α

⋅

1.61 Run 1 0.069 1.61 Run 1 0.075 3.73 Run 1 0.185 2.70 Run 1 0.088 1.90 Run 1 0.103 2.29 Run 1 0.131 8.23 Run 1 0.375 5.10 Run 1 0.144 2.59 Run 1 0.164 3.49 Run 1 0.190 10.3 Run 1 0.446 7.70 Run 1 0.213 3.25 Run 1 0.228 4.34 Run 1 0.228 24.9 Run 1 0.559 9.80 Run 1 0.283 4.60 Run 1 0.292 5.74 Run 1 0.255 1.94 Run 2 0.130 17.8 Run 1 0.363 8.27 Run 1 0.358 8.39 Run 1 0.293 5.34 Run 2 0.260 23.7 Run 1 0.438 10.2 Run 1 0.426 9.94 Run 1 0.327 8.09 Run 2 0.386 60.5 Run 1 0.532 14.9 Run 1 0.489 11.1 Run 1 0.361 15.6 Run 2 0.517 135 Run 1 0.572 45.0 Run 1 0.548 12.0 Run 1 0.394 188 Run 2 0.633 350 Run 1 0.620 168 Run 1 0.617 13.6 Run 1 0.437 326 Run 2 0.664 330 Run 1 0.661 19.3 Run 1 0.475

29.4 Run 1 0.505 53.6 Run 1 0.536 152 Run 1 0.588 381 Run 1 0.645

114

115

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7

1

10

100

1000

10000

293 K - This work - run 1303 K - This work - run 1313 K - This work - run 1313 K - This work - run 2323 K - This work - run 1303 K - Song et al., 2006 313 K - Song et al., 2006 323 K - Song et al., 2006

( )2

-1CO Smol molα ⋅

( )2

210 PaCOP −×

Figure 4.4 – Semi-log plot of the experimental solubility of 2CO in 3.0 M aqueous

solutions of potassium glycinate - comparison with the results from Song et al. (2006)

for an aqueous solution of sodium glycinate 3.09 M, at 303, 313 and 323 K.

Surprisingly, no noticeable difference in the 2CO solubility was observed for

temperatures between 293 and 313 K. This behaviour is unusual in the absorption of

acid-gases in amine based solutions (Benamor and Aroua, 2005; Derks et al., 2005;

Ma'mun et al., 2006; Shen and Li, 1992). Only above 323 K differences start to be

noticeable and at 351 K there is a clear reduction on the solution absorption capacity.

The same trend was observed for the 3.0 M solutions and, for 0.1 M solutions, there were

no differences on the absorption capacity in the entire temperature range studied.

The measured 2CO solubility in 1.0 M potassium glycinate solutions does not differ

significantly from the data obtained by Song et al. (2006) for sodium glycinate 1.06 M,

at 313 and 323 K. Indeed, it should be expected that the absorption capacity of sodium

glycinate is similar to the potassium glycinate aqueous solution used in this work.

Contrarily to the results at 1.0 M, the absorption capacities determined at 3.0 M are not

in line with the results that Song et al. (2006) obtained for sodium glycinate. This can be

due to the difference in the salt cation. Although it is the glycinate anion that reacts with

2CO , the salt cation may start playing a significant role on the process at solution

116

concentrations as high as 3.0 M by modifying the ionic character of the solution.

According to the Guggenheim Equation (17), differences between the short-range

interactions of K + (from potassium glycinate) and Na+ (from sodium glycinate) with

the other ions present in solution would become more noticeable at higher

concentrations.

At high amino acid salt concentrations and high loadings, precipitation of reaction

products was observed by Hook (1997) and Kumar et al. (2003a). The latter authors

concluded that, most likely, the precipitate corresponds to the zwitterionic form of the

amino acid ( 3RNH+ ). They also found a relationship between the critical loading (the

loading at which reaction products start to precipitate), critα , and the solubility of the

zwitterionic form of the amino acid in solution, S :

[ ]2 0

crit

S

RNHα = (24)

Although precipitation was not visually detected during the absorption experiments in

the present work, the critical loading at 3.0 M concentrations was computed using the

solubility data of glycine in water obtained by Ferreira et al. (2004). The minimum

critical loading computed was 0.868 (corresponding to the temperature of 293 K) which

is above the maximum experimental loading at that temperature (0.661). This result

confirms that precipitation was unlikely to occur. Crystallization of the reaction product

3RNH+ would increase the amount of 2CO absorbed (and therefore the solution capacity)

due to the concentration decrease of this reaction product in the liquid phase (Kumar et

al., 2003a).

The effect of the amino acid salt concentration on the absorbent solution capacity at 313

K is shown in Figure 4.5. The loading is shown as a function of the 2CO equilibrium

partial pressure in linear scale. As expected, the loading for a given 2CO partial pressure

decreases with increasing amino acid salt concentration. The same trend was observed

for the other temperatures studied. The 2CO absorption capacity of potassium glycinate

at 313 K was compared to MEA at the same temperature. It was verified that, at that

temperature, potassium glycinate absorbs more than MEA, which is one of the most

widely used 2CO absorbents nowadays.

117

Figure 4.5 - Solution loading as a function of the 2CO equilibrium partial pressure in

aqueous solutions of potassium glycinate at 313 K - comparison with MEA at 2.5 M.

Solid lines are provided to make the figure clearer and do not correspond to theoretical

model results.

2CO solubility in potassium threonate

The absorption capacity of a 1.0 M potassium threonate solution, at 313 K, was also

measured and it is shown in Table 4.5 and Figure 4.6.

Table 4.5 – Experimental solubility of 2CO in aqueous solutions of potassium threonate

1.0 M and 313 K.

( )2

210

Pa

COP −× ( )

2

-1CO Smol mol

α

⋅

1.03 Run 1 0.092 2.28 Run 1 0.188 4.68 Run 1 0.292 8.78 Run 1 0.384 24.5 Run 1 0.479 62.5 Run 1 0.572 190 Run 1 0.674 420 Run 1 0.753

118

Figure 4.6 - Semi-log plot of the experimental solubility of 2CO in aqueous solutions of

potassium threonate and potassium glycinate with concentrations 1.0 M at 313 K.

Comparing the absorption capacity of both amino acid salts at the same solution

concentration, potassium glycinate presents a higher absorption capacity. Threonate has

a deprotonation equilibrium constant, AmAK , higher than glycinate (Perrin, 1965).

Additionally, the amine group is sterically hindered (Portugal et al., 2008), which makes

the carbamate (RNHCOO− ) formed less stable and consequently, carbK is expected to be

higher (although no values were found in literature for this equilibrium constant).

It is generally accepted that amines that form unstable carbamates (high carbK ) present

higher absorption capacities towards 2CO (Baek et al., 2000; Hook, 1997; Li and Chang,

1994; Seo and Hong, 1996). However, this is only true at high 2CO partial pressures

(Park et al., 2003; Sartori and Savage, 1983; Tontiwachwuthikul et al., 1991). On the

other hand, increasing AmAK causes a decrease in loading for the entire pressure range

(Sartori and Savage, 1983).

To clarify how the solution absorption capacities are influenced by the equilibrium

constants AmAK and carbK , the effect of changing these constants by a factor of 10 was

119

checked by simulation, using the Kent and Eisenberg model. The AmAK and carbK values

reported in literature for glycinate (Jensen et al., 1952; Perrin, 1965) – see Table 4.6 –

were used as reference. Results are shown in Figures 4.7 and 4.8.

Figure 4.7 – Effect of changing the carbamate hydrolysis and amine deprotonation

equilibrium constants independently on the predicted 2COP versus loading curves.

Figure 4.8 – Effect of changing the carbamate hydrolysis and amine deprotonation

equilibrium constants simultaneously on the predicted 2COP versus loading curves.

120

The simulated results confirm that increasing carbK alone leads to a mixed effect,

depending on the pressure range considered, as seen in Figure 4.7. On the other hand, the

simultaneous increase of both equilibrium constants, AmAK and carbK , causes a decrease

in loading for the entire pressure range, which is in agreement with what is

experimentally observed in the present work. A more complete analysis would require

knowledge of the carbamate hydrolysis equilibrium constant for threonate. Additionally,

ionic interactions must be considered in these systems and, since threonate and glycinate

anions are different in size and configuration, it is expectable that these interactions will

be different in each case.

Modelling

To model the 2CO absorption equilibrium in potassium glycinate solution, using the

Deshmukh-Mather method (Deshmukh and Mather, 1981), equilibrium constants of

reactions (1) to (5) and the Henry coefficient of 2CO in solutions need to be defined.

Table 4.6 shows the values of the equilibrium constants and Henry coefficient and

respective literature sources.

Table 4.6 – Equilibrium constants of reactions (1) to (5) and Henry coefficient of 2CO in

potassium glycinate solutions.

[ ]2

2

2 0,

log CO

CO w

HK RNH

H

=

62.1830980.111175K

T= − ( )-1M Portugal et al., 2007

( )2 , 7

exp 2044

3.54 10CO w

TH −

−=

× ( )-1 3Pa mol m Versteeg and Van Swaaij, 1988

2767.18exp 6.10312carbK

T

− = +

( )M Jensen et al., 1952

( )2exp 0.000237956 0.202203 61.6499AmAK T T= − + − ( )M Perrin, 1965

2

12092.1exp 36.7816ln 235.482COK T

T = − − +

( )M Benamor and Aroua, 2005

3

12431.7exp 35.4819ln 220.067

HCOK T

T−

= − − +

( )M Benamor and Aroua, 2005

13445.9exp 2.4773ln 140.932wK T

T = − − +

( )2M Benamor and Aroua, 2005

121

Kielland (1937) summarized the effective size of a set of hydrated ions, including the

ions existent in the studied system, except for the carbamate of glycine. Values of

parameter a , based on Kielland’s work are shown in Table 4.7.

Table 4.7 – Effective size of the hydrated ions, based on the work by Kielland (1937).

Ion H + OH − 3HCO− 23CO − 2RNH K + RNHCOO−

Parameter a 9 3.5 4 4.5 4.5 3

5 (estimated)

Kumar et al. (2003b) noticed that the ionic strength of the amino acid salt solutions do

not change significantly during 2CO absorption. For this reason, the long-range

interactions, KiLR , between ions will depend essentially on the solution initial

concentration - [ ]2 0I RNH≃ - and can be directly computed. Concerning the short-range

binary interactions, in reality, when substituting the activity coefficients on the

calculation of the equilibrium constants, only six independent parameters can be fitted:

[ ] ( )2 3

1 3 2exp 2carb Kcarb

RNH HCOK LR p RNH p K

RNHCOO

−+ +

−

= +

(25)

[ ] [ ]

( )3 3 4 5 22

3 5 1

exp 2AmA KAmA

p RNH p K p RNHRNH HK LR

RNH p p RNHCOO

+ ++

+ −

+ − = − −

(26)

[ ] ( )2

3

2 62

exp 2CO KCO

HCO HK LR p K

CO

− ++

= (27)

( )3

23

3 6

3

exp 2KHCOHCO

CO HK LR p K

HCO−

− ++

−

= − (28)

w KwK OH H LR− + = (29)

where the model parameters 1p to 6p are defined as:

122

( )( )( )( )

3 2 3

2 3

3 2 3 3

2 3

3 2

3

1 , ,

2 , , ,

3 , ,

4 , ,

5 ,

6 ,

RNH RNH RNH RNHCOO

RNH K HCO K RNHCOO K

RNH RNH RNH RNH

RNH K RNH K

RNH RNH

HCO K

p

p

p

p

p

p

β β

β β β

β β

β β

β

β

+ + −

+ − + − +

+ + +

+ + +

+

− +

= −

= + −

= −

= −

=

=

(30)

For each 2COP , the set of eight equations (1 vapour-liquid equilibrium equation, 5

reaction equilibrium constants and 2 mass balances) can be reduced to only 1 equation in

terms of the hydrogen ion concentration, H + :

[ ] [ ]

( )

( ) ( )

[ ] ( )( )

2

2 2

2 2 22 2 2 3

2 2 2

2 2 2

2 02 '0

2

' '

' '''

2

'2 0

'2 ' '

'

1

2

20

COCO CO

AmA carb

CO CO COCO CO CO HCOW

CO CO CO

carbcarb CO CO CO

AmA

RNH HRNH H

KH H P H

K K

K K P HK P HK

H H H

RNH K P H

KH K H K P H

K

−

++

+ +

+ + +

+ +

+ − + +

− − −

− = + +

(31)

where 'iK are the equilibrium constants expressed in terms of molar concentrations -

' ii

Ki Ki

KK

LR SR=

⋅ and KiSR denote the short-range interactions term. The correspondent

loading, modα , is computed using equation (14).

A global fit to all temperatures and concentrations was performed. The objective

function to be minimized was:

exp mod

expobjF

α αα−

=∑ (32)

123

The fitted parameters are presented in Table 4.8.

Table 4.8 – Model parameters fitted for the system potassium glycinate-water- 2CO

1p 2p 3p 4p 5p 6p

0.230 -0.877 0.0115 0.351 0.419 0.125

Unlike other amines or amino acids, the amine deprotonation equilibrium constant of

glycine, AmAK , found in literature seem to be very established and several authors

determined it with agreeing results – results summarized by Perrin (1965). However, the

equilibrium constant for the carbamate hydrolysis, carbK , available in the literature is

rather old (Jensen et al., 1952). The simple approach from Kent and Eisenberg (1976)

was also fit considering that:

0ln carb

AK K

T= + (33)

The same data range and objective function were used. The values of 0K and A

obtained were, respectively, -4.786 and 1792 K-1.

Results obtained with both models for the temperature of 293 K are presented in Figure

4.9 and Figure 4.10 shows the parity plot of the predicted and experimental loadings of

2CO in solution for all analysed data.

Figure 4.11 shows the species concentrations as a function of loading, obtained with the

Deskmukh-Mather model, for a 1.0 M solution at 313 K as a function of the loading.

Similar trends were obtained for all studied temperatures and initial amino acid salt

concentrations.

124

Figure 4.9 – Solubility of 2CO in potassium glycinate solutions at 293 K – experimental

values and model curves.

Figure 4.10 – Parity plot of the predicted and experimental loadings of 2CO in solution

for all data analysed.

125

Figure 4.11 – Species concentrations as a function of loading for a potassium glycinate

solution, 1.0 M, at 313 K obtained using the Deskmukh-Mather model. Note that points

are not experimental data but simulation results.

The average relative deviations presented by the Deskmukh-Mather (D&M) and the

Kent and Eisenberg (K&E) models are respectively 22 and 20 %. The error distribution

is shown in Figure 4.10. Apparently, the D&M is able to describe better the experimental

trends, although the K&E shows a lower average deviation. Globally, both models show

similar accuracies which is surprising since the D&M has 6 fitting parameters against the

2 from the K&E model.

Even though the relative deviations may be considered acceptable, Figure 4.9 clearly

shows that both fits can still be considerably improved. Several authors (Benamor and

Aroua, 2005; Ermatchkov et al., 2006; Ma'mun et al., 2006) considered in their

regressions various interaction parameters that were neglected in the present work

(including interactions with 2CO and 23CO − ). The species concentrations as a function of

loading shown in Figure 4.11 suggests that, in fact, 23CO − can be playing a role in the

absorption process and that 2CO concentration can become significant for very high

loadings. However, their concentrations are relatively low, for the loadings considered,

126

when compared to the other species in solution. Benamor and Aroua (2005) and

Ermatchkov et al. (2006) also suggest that the short-range interactions between ions are

temperature dependent. Nevertheless, regressing more than 6 parameters from the

presented data set would lead to unreliable results. For this reason, obtaining the

carbamate hydrolysis equilibrium constant, carbK , as well as some of the binary

interaction parameters, ,i jβ , independently would be a way to improve the prediction

results. Experiments at higher temperatures are also recommended.

4.5. Conclusions

The solubility of 2CO in potassium glycinate and potassium threonate solutions was

measured. The amino acid salts shown absorption capacities in the same order of

magnitude as MEA.

At moderately low temperatures – between 293 and 323 K – no difference was noticed in

the 2CO solubility at different temperatures. However, increasing temperature to about

351 K, the 2CO solubility decreases considerably. Experimental data at higher

temperatures will be very important to understand the dependence of 2CO solubility on

temperature. The temperature at which the solution needs to be heated to efficiently

desorb 2CO will define the amount of energy required for absorbent regeneration, which

will be determinant for the economical viability of the global process.

As observed for other amine based compounds (Benamor and Aroua, 2005; Kumar et al.,

2003b; Song et al., 2006), 2CO solubility in potassium glycinate solutions (expressed in

terms of loading) decreases with increasing potassium glycinate concentration.

2CO solubility in a 1.0 M potassium threonate solution at 313 K was also measured and

compared to potassium glycinate. The trend observed experimentally was qualitatively

confirmed by simulation. However, for a quantitative analysis, the carbamate hydrolysis

equilibrium constant of threonate needs to be determined and binary interactions

between ions should be considered.

127

2CO solubility in the potassium glycinate solutions was interpreted using the Deshmukh-

Mather thermodynamically sound model and the empirical Kent-Eisenberg model.

Although the average deviations between predicted and experimental loadings are lower

than 22 % for both models, the predictions can still be significantly improved. With this

purpose, it is suggested to determine independently the carbamate hydrolysis equilibrium

constant, carbK , and some of the binary interaction parameters, ,i jβ .

4.6. Nomenclature

[ ] Concentration, M

A Debye-Hückel limiting slope

a Parameter corresponding to the effective size of hydrated ions, Å

wa Water activity

B Parameter of equation (17)

H Henry coefficient, -1 3Pa mol m⋅ ⋅


k Experimental stage

K Equilibrium constants, expressed in terms of molarity

LR Long-range interactions

n Number of moles, mol

p Model parameters


R Universal gas constant, -1 -1J mol K⋅ ⋅

S Solubility, M

SR Short-range interactions

T Temperature, K

V Volume, m3

z Ion charge

Greek symbols

α Loading, mol

CO2⋅ mol

S-1

,i jβ Short range interaction parameters

128

ε Dielectric constant of water, K-1

γ Activity coefficient

Subscripts

0 Initial

abs Absorbed

add Added

AV Absorbent vessel

eq Equilibrium

Exp Experimental

fin Final

GV Gas vessel

Mod Model

S Amino acid salt

SC Stirred cell

sol Solution

w Water

4.7. References

Aroua, M. K. and Salleh, R. M. (2004). "Solubility of CO2 in aqueous piperazine and its

modeling using the Kent-Eisenberg approach." Chemical Engineering & Technology,

27(1), 65-70.

Austgen, D. M., Rochelle, G. T., Peng, X. and Chen, C. C. (1989). "Model of Vapor

Liquid Equilibria for Aqueous Acid Gas Alkanolamine Systems Using the Electrolyte

NRTL Equation." Industrial & Engineering Chemistry Research, 28(7), 1060-1073.

Baek, J.-I., Yoon, J.-H. and Eum, H.-M. (2000). "Prediction of equilibrium solubility of

carbon dioxide in aqueous 2-amino-2-methyl-1,3-propanediol solutions." Korean Journal

of Chemical Engineering, 17(4), 484-487.

Benamor, A. and Aroua, M. K., (2005). "Modeling of CO2 solubility and carbamate

concentration in DEA, MDEA and their mixtures using the Deshmukh-Mather model."

Fluid Phase Equilibria, 231(2), 150-162.

129

Chen, C. C. and Evans, L. B., (1986). "A Local Composition Model for the Excess Gibbs

Energy of Aqueous-Electrolyte Systems." American Institute of Chemical Engineers

Journal, 32(3), 444-454.

Clegg, S. L. and Pitzer, K. S. (1992). "Thermodynamics of Multicomponent, Miscible,

Ionic-Solutions - Generalized Equations for Symmetrical Electrolytes." Journal of

Physical Chemistry, 96(8), 3513-3520.

Clegg, S. L., Pitzer, K. S. and Brimblecombe, P. (1992). "Thermodynamics of

multicomponent, miscible, ionic-solutions .2. Mixtures including unsymmetrical

electrolytes." Journal of Physical Chemistry, 96(23), 9470-9479.

Derks, P. W. J., Dijkstra, H. B. S., et al., (2005). "Solubility of carbon dioxide in aqueous

piperazine solutions." American Institute of Chemical Engineers Journal, 51(8), 2311-

2327.

Deshmukh, R. D. and Mather, A. E., (1981). "A Mathematical-Model for Equilibrium

Solubility of Hydrogen-Sulfide and Carbon-Dioxide in Aqueous Alkanolamine

Solutions." Chemical Engineering Science, 36(2), 355-362.

EC (2008). "Communication from the Commission to the European Parliament, The

Council, The European Economic and Social Committee and The Committee of the

Regions - 20 20 by 2020 Europe's climate change opportunity." E. Commission,

Commission of the European Communities.

Ermatchkov, V., Kamps, A. P. S., et al., (2006). "Solubility of carbon dioxide in aqueous

solutions of piperazine in the low gas loading region." Journal of Chemical and


Favre, E., (2007). "Carbon dioxide recovery from post-combustion processes: Can gas

permeation membranes compete with absorption?" Journal of Membrane Science, 294(1-

2), 50-59.

Feron, P. H. M. and Jansen, A. E. (2002). "CO2 separation with polyolefin membrane

contactors and dedicated absorption liquids: performances and prospects." Separation

and Purification Technology, 27(3), 231-242.

130

Ferreira, L. A., Macedo, E. A., et al., (2004). "Solubility of amino acids and diglycine in

aqueous-alkanol solutions." Chemical Engineering Science, 59(15), 3117-3124.

Furst, W. and Renon, H. (1993). "Representation of Excess Properties of Electrolyte-

Solutions Using a New Equation of State." American Institute of Chemical Engineers

Journal, 39(2), 335-343.

Gibbins, J. and Chalmers, H., (2008). "Preparing for global rollout: A 'developed country

first' demonstration programme for rapid CCS deployment." Energy Policy, 36(2), 501-

507.

Goff, G. S. and Rochelle, G. T., (2006). "Oxidation inhibitors for copper and iron



Guggenheim, E. A., (1935). "The specific thermodynamic properties of aqueous

solutions of strong electrolytes." Philosophical Magazine, 19(127), 588-643.

HajiSulaiman, M. Z., Aroua, M. K. and Pervez, M. I. (1996). "Equilibrium concentration

profiles of species in CO2-alkanolamine-water systems." Gas Separation & Purification,

10(1), 13-18.

Holst, J., Politiek, P. P., et al. (2006). "CO2 capture from flue gas using amino acid salt

solutions." GHGT-8, NTNU VIDERE, Pav. A, Dragvoll, NO-7491 Trondheim, Norway.

Hook, R. J., (1997). "An investigation of some sterically hindered amines as potential


36(5), 1779-1790.

Huttenhuis, P. J. G., Agrawal, N. J., Solbraa, E. and Versteeg, G. F. (2008). "The

solubility of carbon dioxide in aqueous N-methyldiethanolamine solutions." Fluid Phase

Equilibria, 264(1-2), 99-112.

Idem, R. and Tontiwachwuthikul, P., (2006). "Preface for the special issue on the capture

of carbon dioxide from industrial sources: Technological developments and future

opportunities." Industrial & Engineering Chemistry Research, 45(8), 2413-2413.

IEA (2008). "International Energy Agency - http://www.iea.org/."

131

Jensen, A., Jensen, J. B., et al., (1952). "Studies on Carbamates .6. The Carbamate of

Glycine." Acta Chemica Scandinavica, 6(3), 395-397.

Jones, J. H., Froning, H. R., et al., (1959). "Solubility of Acidic Gases in Aqueous

Monoethanolamine." Journal of Chemical Engineering Data, 4(1), 85.

Kielland, J. (1937). "Individual activity coefficients of ions in aqueous solutions."

Journal of the American Chemical Society, 59, 1675-1678.

Kent, R. L. and Eisenberg, B., (1976). "Better Data for Amine Treating." Hydrocarbon

Processing, 55(2), 87-90.

Knowlton, A. E., (1941). "Standard handbook for electrical engineers" McGraw-Hill,

New York.

Kumar, P., Hogendoorn, J., et al., (2002). "New absorption liquids for the removal of

CO2 from dilute gas streams using membrane contactors." Chemical Engineering

Science, 57(9), 1639-1651.

Kumar, P., Hogendoorn, J., et al., (2003a). "Equilibrium solubility of CO2 in aqueous

potassium taurate solutions: Part 1. Crystallization in carbon dioxide loaded aqueous salt

solutions of amino acids." Industrial & Engineering Chemistry Research, 42(12), 2832-

2840.

Kumar, P., Hogendoorn, J., et al., (2003b). "Equilibrium solubility of CO2 in aqueous

potassium taurate solutions: Part 2. Experimental VLE data and model." Industrial &


Kumar, P., Hogendoorn, J., et al., (2003c). "Kinetics of the reaction of CO2 with aqueous

potassium salt of taurine and glycine." American Institute of Chemical Engineers

Journal, 49(1), 203-213.

Lee, J. I., Otto, F. D., et al., (1974). "Solubility of Mixtures of Carbon-Dioxide and

Hydrogen-Sulfide in Aqueous Diethanolamine Solutions." Canadian Journal of Chemical

Engineering, 52(1), 125-127.

132

Lee, J. I., Otto, F. D., et al., (1976). "Equilibrium in Hydrogen Sulfide

Monoethanolamine-Water System." Journal of Chemical and Engineering Data, 21(2),

207-208.

Li, C. X. and Furst, W. (2000). "Representation of CO2 and H2S solubility in aqueous

MDEA solutions using an electrolyte equation of state." Chemical Engineering Science,

55(15), 2975-2988.

Li, M. H. and Shen, K. P. (1993). "Calculation of Equilibrium Solubility of Carbon-

Dioxide in Aqueous Mixtures of Monoethanolamine with Methyldiethanolamine." Fluid

Phase Equilibria, 85, 129-140.

Li, M. H. and Chang, B. C. (1994). "Solubilities of Carbon-Dioxide in Water Plus

Monoethanolamine Plus 2-Amino-2-Methyl-1-Propanol." Journal of Chemical and


Liu, H. B., Zhang, C. F. and Xu, G. W. (1999). "A study on equilibrium solubility for

carbon dioxide in methyldiethanolamine-piperazine-water solution." Industrial &


Ma'mun, S., Jakobsen, J. P., Svendsen, H. F. and Juliussen, O. (2006). "Experimental

and modeling study of the solubility of carbon dioxide in aqueous 30 mass % 2-((2-

aminoethyl)amino)ethanol solution." Industrial & Engineering Chemistry Research,

45(8), 2505-2512.

Metz, B., Davidson, O., et al. (2005). "IPCC Special Report on Carbon Dioxide Capture

and Storage." C. U. Press. Cambridge, New York, Melbourne, Madrid, Cape Town,

Singapore, São Paulo, Intergovernmental Panel on Climate Change.

Park, J. Y., Yoon, S. J. and Lee, H. (2003). "Effect of steric hindrance on carbon dioxide

absorption into new amine solutions: Thermodynamic and spectroscopic verification

through solubility and NMR analysis." Environmental Science & Technology, 37(8),

1670-1675.

Park, S. H., Lee, K. B., Hyun, J. C. and Kim, S. H. (2002). "Correlation and prediction of

the solubility of carbon dioxide in aqueous alkanolamine and mixed alkanolamine

solutions." Industrial & Engineering Chemistry Research, 41(6), 1658-1665.

133

Perrin, D. (1965). "Dissociation Constants of Organic Bases in Aqueous Solutions."

Butterworth, London.

Portugal, A. F., Derks, P. W. J., et al., (2007). "Characterization of potassium glycinate

for carbon dioxide absorption purposes." Chemical Engineering Science, 62(23), 6534-

6547.

Portugal, A. F., Magalhães, F. D., et al., (2008). "Carbon dioxide absorption kinetics in

potassium threonate." Chemical Engineering Science, 63(13), 3493-3503.

Rascol, E., Meyer, M. and Prevost, M. (1996). "Simulation and parameter sensitivity

analysis of acid gas absorption into mixed alkanolamine solutions." Computers &

Chemical Engineering, 20, S1401-S1406.



Seo, D. J. and Hong, W. H. (1996). "Solubilities of carbon dioxide in aqueous mixtures

of diethanolamine and 2-amino-2-methyl-1-propanol." Journal of Chemical and


Shen, K. P. and Li, M. H., (1992). "Solubility of Carbon-Dioxide in Aqueous Mixtures

of Monoethanolamine with Methyldiethanolamine." Journal of Chemical and


Smith, J. M., Ness, H. C. V., et al., (1996). "Introduction to chemical engineering

thermodynamics." McGraw-Hill, Singapore.

Song, H. J., Lee, S., et al., (2006). "Solubilities of carbon dioxide in aqueous solutions of

sodium glycinate." Fluid Phase Equilibria, 246(1-2), 1-5.

Supap, T., Idem, R., et al., (2006). "Analysis of monoethanolamine and its oxidative



combinations." Industrial & Engineering Chemistry Research, 45(8), 2437-2451.

134

Tobiesen, F. A., Juliussen, O. and Svendsen, H. F. (2008). "Experimental validation of a

rigorous desorber model for CO2 post-combustion capture." Chemical Engineering

Science, 63(10), 2641-2656.

Tontiwachwuthikul, P., Meisen, A. and Lim, C. J. (1991). "Solubility of CO2 in 2-

Amino-2-Methyl-1-Propanol Solutions." Journal of Chemical and Engineering Data,

36(1), 130-133.

UNFCCC (2008). "United Nations Framework Convention on Climate Change -

http://unfccc.int. "

Vallee, G., Mougin, P., Jullian, S. and Furst, W. (1999). "Representation of CO2 and

H2S absorption by aqueous solutions of diethanolamine using an electrolyte equation of

state." Industrial & Engineering Chemistry Research, 38(9), 3473-3480.

Versteeg, G. F. and Van Swaaij, W. P. M., (1988). "Solubility and Diffusivity of Acid



Weiland, R. H., Chakravarty, T., et al., (1993). "Solubility of Carbon-Dioxide and

Hydrogen-Sulfide in Aqueous Alkanolamines. " Industrial & Engineering Chemistry

Research, 32(7), 1419-1430.

Part IV

137

5. Carbon dioxide removal from

anaesthetic gas circuits using hollow

fiber membrane contactors with amino

acid salt solutions 1

Abstract

A novel technology, based on the use of hollow fiber membrane contactors with

regenerable liquid absorbents is proposed for the carbon dioxide removal from

anaesthetic closed breathing circuits. To analyse the performance of the contactor for

this specific application and the influence of the system parameters, a 2 D numerical

model was developed for the transport of 2CO through the hollow fibers. The model

considered potassium glycinate solutions as absorbents and a composite membrane,

made of a porous support layer and a dense thin layer. Both co- and counter-current

operations were studied. The model results were compared to results obtained with

conventional mass transfer models, valid for limit conditions, and a good agreement

was found. The analysis performed indicates that the use of hollow fiber membrane

contactors with amino acid salt absorbent solutions is suitable for 2CO removal from

closed anaesthetic circuits. Contactor design and operating conditions are suggested.

1 Portugal, A. F.; Magalhães, F. D.; Mendes, A., “Carbon dioxide removal from anaesthetic gas circuits using hollow fiber membrane contactors with amino acid salt solutions”, submitted to J. Memb. Sci.

138

5.1. Introduction

Low flow or closed loop anaesthesia is a current clinical practice that consists in feeding

back to the patient (in the subsequent inhalation) the unused anaesthetic gas stream

(Baum and Woehlck, 2003). In such systems, 2CO needs to be continuously removed

from the breathing circuit. For this purpose, mixtures of alkali hydroxides are

commonly used as absorbents (Baum and Woehlck, 2003). However, all halogenated

volatile anaesthetics react with conventional 2CO absorbents when these become

accidentally desiccated, resulting in toxic compounds such as carbon monoxide and the

so called compound A (Baum and Woehlck, 2003; Fan et al., 2008; Knolle and Gilly,

2000; Whalen et al., 2005). In addition, the spent absorbents are contaminated hospital

waste, therefore requiring specific and expensive treatments (Mendes, 2000).

Hollow fiber membrane contactors with renewable liquid absorbents can be an

attractive technology to perform the 2CO removal from respiratory circuits in

anaesthesia machines, as sketched in Figure 5.1 (Mendes, 2000; Portugal et al., 2007).

Because of the absence of interpenetration of the gaseous and liquid phases, membrane

contactors overcome a number of operational limitations, common in other sorts of

contactors, and enable aseptic operation, making the process suitable for the required

application (Gabelman and Hwang, 1999; Li and Chen, 2005).

In absorbent membrane contactors, the selectivity is mostly provided by the liquid and

the driving force for the mass transfer is the concentration gradient between gas and

liquid phases (Gabelman and Hwang, 1999; Li and Chen, 2005). Therefore, the

membrane works as a phase separator and should impose the least possible resistance to

mass transfer. Hence, porous membranes with gas filled pores are usually used (Boucif

et al., 2008; Feron and Jansen, 2002; Sea et al., 2002; Zhang and Cussler, 1985).

However, for the proposed application, the membranes must also block the possible

passage of pathogenic microorganisms from the breathing circuit to the absorbent

solution and the subsequent transmission to the next patient under surgery. A possible

strategy to meet this demand is the use of supported dense polymer membranes as

suggested by Kreulen et al. (1993b) for other applications. Although the dense coating

139

imposes an extra resistance to mass transfer, it also prevents the pores from wetting,

independently of the liquid surface tension and contact angle (Kreulen et al., 1993b).

Figure 5.1 – Sketch of a closed anaesthetic breathing circuit using hollow fiber

membrane contactors for 2CO removal.

Chemical absorption in reactive liquids (usually aqueous solutions of alkanolamines) is

an established technology to perform 2CO separation from a variety of gas mixtures

(Idem and Tontiwachwuthikul, 2006; Ma'mun et al., 2007). However, alkanolamines

may undergo oxidation in environments with high oxygen concentrations (Goff and

Rochelle, 2006; Hook, 1997; Supap et al., 2006). Therefore other absorbent systems are

being developed, namely aqueous solutions of amino acid salts (Feron and Jansen,

2002; Hamborg et al., 2007; Kumar et al., 2002; Portugal et al., 2008; Song et al.,

2006). These solutions are much more resistant to oxidative degradation, are more

thermally stable, present lower volatilities and higher surface tensions and have

densities and viscosities similar to water. Besides, when compared to alkanolamines,

they present similar reaction equilibrium and kinetics (Feron and Jansen, 2002;

Hamborg et al., 2007; Kumar et al., 2002; Portugal et al., 2008; Song et al., 2006). An

important drawback may be related to precipitation of the reaction products, which has

been reported for high amino acid salt concentrations and high loadings (Hook, 1997;

Kumar et al., 2003c; Majchrowicz et al., 2006). This can result in the blockage of the

140

membrane pores (if porous membranes are to be used) and originate hydrodynamic

problems when small fiber diameters are used.

In the present work, a 2D numerical model is developed to describe the 2CO absorption

in amino acid salt solutions. A membrane contactor is considered, with the absorbent

flowing in the hollow fibers bore, and with the gas flowing in the shell side. Both co-

and counter-current operations are considered. To validate the model, simulation results

are compared to results obtained by conventional mass transfer models, at limit

conditions where these are applicable.

For the anaesthetic closed loop simulations, the permeability properties of a composite

PDMS (polydimethylsiloxane) membrane and of aqueous absorbent solutions of

potassium glycinate were used. Specific aspects of this system such as the extra

resistance imposed by the membrane coating, the reaction mechanism and the

absorption equilibrium of 2CO in potassium glycinate are taken into account in the

simulations. The performance of the contactor is analysed and the influence of the

packing density, fiber length, liquid flow rate and solution concentration on the

separation achieved is evaluated.

5.2. Mass Transfer with Chemical Reaction

5.2.1. Chemical reaction

Alkali salts of primary amino acids present identical reactions towards 2CO as primary

alkanolamines (Kumar et al., 2003a). 2CO reacts with aqueous solutions of primary

alkanolamines forming carbamates, bicarbonates and carbonates (Caplow, 1968) and

the first reaction taking place is the formation of the carbamate or an intermediate

(Hook, 1997; Sartori and Savage, 1983):

2 2 32RNH CO RNHCOO RNH− ++ + (1)

It is generally accepted that reaction (1) occurs according to the zwitterion mechanism

(Caplow, 1968). Zwitterion mechanism considers the following reaction steps:

141

Formation of the zwitterion

2

12 2 2

k

kRNH CO RNH COO

−

+ −→+ ← (2)

Deprotonation of the zwitterion by a base

2B

B

k

i ikRNH COO B RNHCOO B H

−

+ − − +→+ +← (3)

where iB are the bases present in solution able to deprotonate the zwitterion, which

includes the amine itself (Blauwhoff et al., 1984; Portugal et al., 2007). Based on the

quasi-steady-state condition for the zwitterion concentration, the following expression

for the rate of 2CO absorption can be written:

2 2

2

1

2

1

2 2

1

i i

i i

i i

B B Hi

CO RNH RNHCOOB B

iCO

B Bi

k Ck

C C Ck k C

Sk

k k k C

+

−

−−

−

−

=+

∑

∑

∑

(4)

Considering that the amine is playing the main role on the zwitterion deprotonation, the

kinetic constants of reactions (2) and (3) can be related to the equilibrium constant of

reaction (1), ovK , as follows:

2

1

i

i

Bov

B

kkK

k k− −

= (5)

Substituting equation (5) in (4) and since, for primary amines, the deprotonation of the

zwiteterion is relatively fast when compared to the backward rate of reaction (2)

( 1 1i iB B

i

k

k c− <<

∑), the following expression for the rate of 2CO absorption results:

3

2 2 2

2

22

RNHCOO RNHCO CO RNH

ov RNH

C CkS k C C

K C

− +

= − (6)

And, according to reaction (1), the rate of consumption of the amino acid is given by:

2 2

2RNH COS S= (7)

142

5.2.2. Analogy to conventional mass transfer models

The process of mass transfer of a gaseous solute A in a membrane contactor includes

the following steps: diffusion from the gas bulk to the outer membrane surface,

diffusion trough the membrane pores, diffusion and sorption in the dense layer (if

coated membranes are used) and dissolution and diffusion in the liquid accompanied (or

not) by chemical reaction (Li and Chen, 2005). Then, the local flux of A into the liquid,

AJ , can be expressed by the following equation:

A ov AJ k C= ∆ (8)

where AC∆ is the concentration gradient between gas and liquid phases and ovk is the

overall mass transfer coefficient that, relating to the resistance in series model, is given

as follows:

liquid phase resistencemembrane resistencegas phase resistence

1 1 1 1

ov g m Lk k k mk E= + + (9)

Usually, the gas phase mass transfer resistance needs to be obtained empirically and is

very specific of the type of module used (Gabelman and Hwang, 1999). If porous non-

wetted membranes are used, the mass transfer resistance due to the membrane is usually

negligible. However, when using coated membranes, membrane resistance needs to be

accounted for. Even in these cases, generally, the liquid phase mass transfer resistance

controls the process.

There are a number of mass transfer models to describe the absorption of a gas into a

liquid under well-mixed bulk conditions (Danckwerts, 1970). Since the flow along a

hollow fiber is usually laminar (given its small diameter) there is a velocity profile

across the entire fiber radius. Therefore, a well-mixed bulk cannot be considered

(Dindore et al., 2005a; Kreulen et al., 1993b; Kumar, 2002). However, there are limiting

situations where the analogy to conventional mass transfer models is possible (Elk et al.,

2007; Knaap et al., 2000; Kreulen et al., 1993a; Kreulen et al., 1993b; Kumar et al.,

2003d).

143

Concerning the physical absorption of a gas with constant composition into a liquid, the

following applies (Dindore et al., 2005a):

( ), ,A L A int A bulkJ k C C= − (10)

where ,A intC is the interfacial concentration and, ,A bulkC is the concentration of A in the

liquid bulk.

Making the analogy to heat transfer models, Kreulen et al. (1993a) proposed an

approximate solution to compute the mass transfer coefficient, Lk , when the liquid is

flowing laminarly (with fully developed velocity profile) along the fiber lumen:

3 3 33.67 1.62Sh Gz= + (11)

where Sh and Gz are the dimensionless Sherwood and Graetz numbers.

For the same flow conditions ,A bulkC can be approximated by the average mixing cup

concentration of A over the length of the fiber, ,A LC . Considering a lean liquid entering

a hollow fiber with diameter d , the mixing cup concentration at the axial distance z

from the liquid inlet, ,A zC , is given by (Kreulen et al., 1993a):

, ,

41 exp L

A z A int

k zC C

v d

= − −

(12)

Integrating ,A zC over the fiber length, the average mixing cup concentration of A is

obtained (Dindore et al., 2005a; Kumar, 2002):

, , ,

4exp 1

4L

A L A int A intL

L kv dC C C

L k v d

= + − −

(13)

where L is the fiber length and v is the liquid velocity.

The approximation proposed by Kreulen et al. (1993a) was experimentally validated

and shown to be applicable over the entire range of laminar flow.

In the case of plug flow, according to the penetration theory, the mass transfer

coefficient is given by (Bird et al., 2002):

2 AL

D vk

Lπ= (14)

144

being AD the diffusion coefficient of the absorbing gaseous component A .

When the gas absorption is accompanied by chemical reaction, an enhancement factor,

E , is introduced in the calculation of the absorption flux (Dindore et al., 2005a):

( ), ,A L A int A bulkJ E k C C= − (15)

Generally, the enhancement factor is a function of the dimensionless Hatta number,

Ha , and the infinite enhancement factor, E∞ , which for a general reaction

ProductsA BA Bν ν+ → , with reaction rate expression ,m n

A m n A BS k C C= , are defined as:

1

, , ,n m

m n B bulk A int A

L

k C C DHa

k

−

= (16)

,

,

1q

B bulk B A

B A int A B

C D DE

C D Dν∞

= +

(17)

being ,m nk the reaction kinetic constant, ,B bulkC the reactant concentration in the liquid

bulk and BD the diffusion coefficient of reactant B in the liquid. The value of q

depends on the mass transfer model chosen, being 0 for the film model, 1/2 for the

penetration model and 1/3 for the Leveque model (which accounts for the presence of a

velocity gradient in the mass transfer zone).

Based on the surface renewal theory, DeCoursey (1974) developed an explicit

expression to calculate the enhancement factor of a second order irreversible reaction

( ProductsA BA Bν ν+ → with 2A A BS k C C= ):

( ) ( )22 4

2 12 1 14 1

E HaHa HaE

E EE∞

∞ ∞∞

= − + + +− −−

(18)

The DeCoursey approximation proved to accurately predict the enhancement factors

over a wide range of process conditions (Van Swaaij and Versteeg, 1992).

According to the film theory, Secor and Beutler (1967) derived an equation for the

infinite enhancement factor, including reversibility, of the following general absorption

reaction:

145

A B C DA B C Dν ν ν ν+ + (19)

The expression from Secor and Beutler (1967) was later adapted by Hogendoorn et al.

(1997) to make it more compatible with the penetration and surface renewal theories

(Derks et al., 2006):

( )( )

, ,

, ,

1 A C int C bulk C

AC A int A bulk

C C DE

DC C

νν∞

−= +

− (20)

The concentration of the product C at the interface, ,C intC can be computed using the

following equations (Derks et al., 2006):

( ), , , ,CB

B int B bulk C bulk C intC B

DC C C C

D

νν

= + − (21)

( ), , , ,CD

D int D bulk C bulk C intC D

DC C C C

D

νν

= − − (22)

, ,

, ,

C D

A B

C int D intov

A int B int

C CK

C C

ν ν

ν ν= (23)

Equations (18) and (20) can then be combined to compute the enhancement factor for

reversible reactions (Hogendoorn et al., 1997).

For short gas-liquid contact times, penetration depth of the gas into the liquid is smaller

than the fiber radius and the liquid can be considered of infinite depth (Kumar et al.,

2003b). For this situation, the centreline concentration of the liquid reactant remains

essentially constant and the analogy to the reported models can be made based on the

fiber inlet conditions (Kumar et al., 2003b), that is , ,B bulk B feedC C≈ .

Although the applicability of the analogies to conventional mass transfer models covers

a wide range of asymptotic conditions, for reactive absorption, small fiber diameters and

large contact times, the depletion of the reactant along the fiber axis might become

significant and a model based on first principles is required to accurately predict the

absorption flux (Dindore et al., 2005a; Kreulen et al., 1993b; Kumar et al., 2003b;

Kumar, 2002). Besides, for unequal diffusivities of the species involved in the reaction,

the approximations tend to deviate from the real solution, even when a well-mixed bulk

146

is present (Hogendoorn et al., 1997). Finally, to assess axial gradients in gas phase

velocity and composition, a coupled differential model for both gas and liquid phases

must be solved.

5.2.3. Mathematical model

Numerous models for the mass transfer in hollow fiber absorbent membrane contactors,

with the liquid flowing in the fibers lumen, have been proposed and solved. Among

these, some consider the governing equations for both gas and liquid phases (Al-

Marzouqi et al., 2008; Coker et al., 1998; Dindore et al., 2005b; Hoff et al., 2004;

Keshavarz et al., 2008; Zhang et al., 2006). Other works, on the other hand, are focused

on the liquid phase, assuming uniform gas velocity and concentration along the

contactor’s axial coordinate (Bao and Trachtenberg, 2005; Boucif et al., 2008; Chen et

al., 2007; Kumar et al., 2002; Li and Chen, 2005; Paul et al., 2007; Wang et al., 2004).

Since, in the present work, one is concerned with the 2CO concentration in the outlet

gas stream, and not just with the global amount of 2CO removed, the model considered

integrates both gas and liquid phases, being 1D for the gas and 2D for the liquid. The

unsteady state model was developed with the following main assumptions: negligible

temperature effects (Kumar, 2002), negligible pressure drop in both shell and lumen and

applicability of the Henry law (Zhang et al., 2006).

Assuming plug flow, ideal gas behaviour and constant feed pressure in the gas phase,

the following mass balances can be written:

Partial gas phase mass balance

( )

, , ,21

iig T g T i m

uyyC f C N

t z

εε

∂∂ = −∂ ∂ −

(24)

Total gas phase mass balance

,21 i m

i

uf N

z

εε

∂ =∂ − ∑ (25)

with initial and boundary conditions:

147

Initial condition

0t = , ,0i iy y=

Axial boundary condition

,

,

co-current: 0, and

counter-current: , and i i feed feed

i i feed feed

z y y u u

z L y y u u

= = =

= = =

where iy is the gas molar fraction of component i , ,g TC is the total molar

concentration, which, for ideal gas behaviour is given by ( ),g TC P RT= , where P is

the operation pressure, R is the ideal gas constant and T is the temperature; u is the

gas velocity, ε is the module packing density, defined as 2 2inner shellnR R , being n the

number of fibers, innerR the fibber inner radius and shellR the shell internal radius; f

assumes the value of 1 or -1, respectively, for counter- or co-current operation. The flux

of the gaseous component i across the membrane, ,i mN , is given by:

,,

, ,inner

i L Ri exti m i g T

inner i

CkN y C

R m

= −

(26)

where and ,i extk is the external mass transfer coefficient, which takes into account the

gas and membrane mass transfer resistances: ,, ,

1

1 1i exti g i m

kk k

=+

, ,inner

i L RC is the

concentration of i in the liquid, at the wall and im is the partition coefficient of i ,

which accounts for the physical solubility.

Concerning the liquid phase, axi-symmetry and negligible axial diffusion were assumed

(Dindore et al., 2005a; Kumar, 2002). Both plug and laminar flows were considered in

the axial direction and diffusive flow is assumed in the radial direction. Then, the

differential mass balance of any absorbing species i present in the liquid phase and the

reactant species B are given by:

Liquid phase mass balance to the absorbed component i

, , ,1i L i L i Li i

C C Cv D r S

t z r r r

∂ ∂ ∂ ∂= − + − ∂ ∂ ∂ ∂ (27)

148

Liquid phase mass balance to the reactive component B

1B B B

B B

C C Cv D r S

t z r r r

∂ ∂ ∂ ∂ = − + − ∂ ∂ ∂ ∂ (28)

with the following initial and boundary conditions:

Initial condition

0t = , , , ,0i L i LC C= and ,0B BC C=


0z = , , , ,i L i L feedC C= and ,B B feedC C=

Axi-symmetry condition

0r = , , 0i LC

r

∂=

∂ and 0BC

r

∂ =∂

Mass transfer across the membrane

innerr R= , ,,

, ,i

i L Ri Li i ext i g

i

CCD k C

r m

∂ = − ∂

(Absorbed gases)

0BC

r

∂ =∂

(Non-volatile reactant)

where ,i LC is the concentration of specie i in the liquid, v is the liquid velocity:

2

2 1i

rv v

R

= −

for laminar velocity profile and v v= for plug flow, being v the

average liquid velocity; iD and BD are the diffusion coefficients of absorbed species i

and reactant species B ; iS and BS are the terms accounting for the chemical reaction.

To put in evidence the governing parameters of the process, model equations are also

presented in their dimensionless form.

Partial gas phase mass balance

( )*

*/ / ,2

1ii

L G L M i m

u yyf R R N

xτ τε

θ ε∂∂ = −

∂ ∂ − (29)

149

Total gas phase mass balance

*

*/,

/

21

L Mi m

iL G

Ruf N

x R

τ

τε

ε∂ =∂ − ∑ (30)

with initial and boundary conditions:

Initial condition

0θ = , ,0i iy y=


*,

*,

co-current: 0, and 1

counter-current: 1, and 1

i i feed

i i feed

x y y u

x y y u

= = =

= = =

where θ is the dimensionless time defined as Lt τ and Lτ is the liquid phase residence

time: L L vτ = ; x is the dimensionless axial coordinate, z L ; *u is the gas velocity

normalized by the gas feed velocity: *feedu u u= ; /L GRτ is the ratio between liquid and

gas residence times, L gτ τ , being the gas phase residence time, gτ , defined as feedL u ;

/L MRτ is the ratio between the liquid residence time and a membrane characteristic time,

mτ , defined as ,i A extR k , where ,A extk is the external mass transfer coefficient for the

reference component A . The dimensionless flux of the gaseous component i across the

membrane, *,i mN , is given by:

*,

* *, , *

inneri L R

i m i ext ii

CN k y

m

= −

(31)

where *,i extk and *

im are, respectively, the external mass transfer coefficient and the

partition coefficient of i normalized by the reference component A .

Liquid phase mass balances can be written in the dimensionless form as follows:

Liquid phase mass balance to the absorbed component i

* * **, , ,* *1

4i L i L i Lii

C C CDv Da S

x Gzρ

θ ρ ρ ρ ∂ ∂ ∂∂= − + − ∂ ∂ ∂ ∂

(32)

150

Liquid phase mass balance to the reactive component B

* * * *

,* *

,

14 L refB B B B

BB feed

CC C D Cv Da S

x Gz Cρ

θ ρ ρ ρ ∂ ∂ ∂∂= − + − ∂ ∂ ∂ ∂

(33)

with the following initial and boundary conditions:

Initial condition

0θ = , * *, , ,0i L i LC C= and * *

,0B BC C=


0x = , * *, , ,i L i L feedC C= and * 1BC =

Axi-symmetry condition

0ρ = , *, 0i LC

ρ∂

=∂

and *

0BC

ρ∂ =∂

Mass transfer across the membrane

1ρ = ,

** *,, , /

* *

Gz1

4i

i L Ri L i ext L Mi

A i i

CC k Ry

m D m

τ

ρ

∂ = − ∂

(Absorbed gases)

*

0BC

ρ∂ =∂

(Non-volatile reactant)

where *,i LC is the dimensionless concentration of specie i in the liquid,

*, , ,i L i L L refC C C= , being the reference liquid concentration,,L refC , defined as ,A g Tm C ; *v

is the dimensionless liquid velocity (*v v v= ); ρ is the dimensionless radial

coordinate, innerr R ; *iD is the diffusion coefficient of component i normalized by the

reference component A : *i i AD D D= . The Graetz dimensionless number, Gz, relates

the convection and diffusion characteristic times and is given by: 2

A

vdGz

D L= . The

Damköhler dimensionless number, Da , which relates convection and reaction

characteristic times, is defined here as 2 ,L B feedDa k Cτ= , being ,B feedC the concentration

of the non-volatile reactant at the entrance of the contactor and 2k the reaction kinetic

constant; *iS is the dimensionless term accounting for the chemical reaction and is equal

to zero for all non-reactive species in solution. For a second order irreversible reaction

with 1A Bν ν= = : * * * *,A B A L BS S C C= = . If the absorption rate expression (6) applies:

151

( )2*

* * *, *

,

11

4B

A A L Bov L ref B

CS C C

K C C

−= − and * *2B AS S= . *

BC is the dimensionless concentration

of reactant B in the liquid, normalized by the feed concentration: *,B B B feedC C C= .

5.2.4. Numerical resolution strategy

Even though one is interested only on the steady state solution, the unsteady state model

was solved in order to overcome numerical instability problems. These are likely to

occur for conditions that lead to steep concentration profiles (high reactant

concentrations and high liquid velocities) and for counter-current operation (Sousa and

Mendes, 2003). The strategy used to solve the resulting system of partial differential

equations, (29) to (33), consisted on the spatial discretization of each equation, using the

finite volumes method (Cruz et al., 2005; Santos et al., 2007), and the subsequent time

integration (method of lines).

Concerning the spatial discretization, both shell and fiber were divided in equally

spaced intervals along the axial direction. The fiber lumen was also divided in the radial

direction following a geometric progression, therefore providing a higher number of

volumes next to the hollow fiber wall, where the concentration profiles are steepest. The

concentrations in each cell face were calculated using a first-order upwind differencing

scheme (Courant et al., 1952), for both axial and radial faces, and the radial derivatives

were computed using a second order central differencing scheme (Santos et al., 2007).

Details on this spatial discretization are presented in Appendix.

The resulting time dependent ordinary differential equations were then integrated using

the time integration package LSODA (Petzold and Hindmarsh, 1997). This routine

solves initial value problems, consisting on stiff or non-stiff systems of first order

ordinary differential equations, with variable step size and convergence order.

The solution is considered to be in steady state when the time derivative of each

dependent variable, for each spatial coordinate, is smaller than a pre-defined value.

152


5.3.1. Model validation

The results obtained with the model described above were compared to those from

conventional mass transfer models, for the limiting situations where the analogy is

valid. To keep the gas phase composition constant, pure 2CO was considered for these

simulations.

Physical absorption flux of 2CO in water, under laminar and plug flow, are shown in

Figure 5.2 as a function of Gz. The results from the conventional approach, were

obtained from equations (10) to (13), for laminar flow, and equation (14), for plug flow.

From this figure it can be seen that the numerical model matches closely with the

simplified models for both flow patterns.

Figure 5.2 – Absorption flux of 2CO in water as a function of Gz – numerical model

(NM) and conventional model (CM) results. Simulation conditions: 0.196ε = ,

7/ 5.818 10L MGz Rτ⋅ = × , 0.833Am = , -3

, 40.34 mol mg TC = ⋅ .

153

The absorption in the presence of chemical reaction was studied for the limiting

condition where the depletion of the reactant at the fiber axis is negligible. Three typical

reaction types were considered in the simulations:

2kA B C D+ → + with 2A A BS k C C= (34)

2

1

k

kA B C D

−

→+ +← with 2 1A A B C DS k C C k C C−= − , 2

1

C Dov

A B

C CkK

k C C−

= = (35)

2

1

2k

kA B C D

−

→+ +← with 2 1C D

A A BB

C CS k C C k

C−= − , 22

1

C Dov

A B

C CkK

k C C−

= = (36)

In Figure 5.3, the results from the numerical model (NM) and conventional model (CM)

are plotted in terms of E as a function of Ha . Values of Ha were varied by changing

the reaction kinetic constant, 2k . The conventional approach results were obtained using

the DeCoursey equation (18). The physical mass transfer coefficients used to compute

Ha - equation (16) - were obtained from equations (11) and (14), respectively for

laminar and plug flow. The infinite enhancement factors, E∞ , used for reaction (34) was

calculated using equation (17) and the enhanced factors of reactions (35) and (36) were

computed using equations (20) to (23).

Figures 5.2 and 5.3 show that simulations results obtained with the model developed in

this work are in agreement with results from conventional mass transfer models, with

slight differences (especially in the intermediate regime), inherent to the approximations

involved in the conventional models (Kumar et al., 2003b). This validates the developed

model and the numerical strategy adopted.

154

Figure 5.3 – E vs Ha plot for reactions (34), (35) and (36) – numerical (NM) and

conventional (CM) models results. Simulation conditions: 412.63Gz= , / 19.795L GRτ = ,

5/ 1.36 10L MRτ = × , 0.196ε = , 1Am = , * 1BD = , -3

, 41.6 mol mg TC = ⋅ , , 5MB feedC = and

0.8K = .

Radial concentration profiles of the reactant and the absorbed gas at the exit of the

contactor, obtained using the numerical model, are plotted in Figures 5.4 and 5.5 for

3253Ha = and 3.253Ha = for laminar flow - these conditions correspond, respectively

to instantaneous (IR) and fast pseudo-first order reaction (PFO) regimes (Danckwerts,

1970). The profile for physical absorption in water is also shown for comparison. Figure

5.4 states for a direct reaction type (34) and Figure 5.5 for a reversible reaction type (35)

.

155

Figure 5.4 – Radial profiles at the fiber outlet for pseudo first order (PFO) and

instantaneous reaction (IR) regimes, for a direct second order reaction – equation (34).

Simulation conditions: Laminar flow, 412.63Gz= , / 19.795L GRτ = , 5/ 1.36 10L MRτ = × ,

0.196ε = , 1Am = , * 1BD = , -3, 41.6 mol mg TC = ⋅ , , 5MB feedC = .

Figure 5.5 – Radial profiles at the fiber outlet for pseudo first order (PFO) and

instantaneous reaction (IR) regimes, for a direct second order reaction – equation (35).

Simulation conditions: Laminar flow, 412.63Gz= , / 19.795L GRτ = , 5/ 1.36 10L MRτ = × ,

0.196ε = , 1Am = , * 1BD = , -3, 41.6 mol mg TC = ⋅ , , 5MB feedC = and 0.8K = .

156

Figures 5.4 and 5.5 make clear that, at these conditions, the concentrations of both

reactant and absorbed gas stay practically unchanged along the fiber axis. Figure 5.4

also shows that, as expected, under pseudo first order reaction regime, the depletion of

the reactant at the fiber wall in negligible, while under the instantaneous reaction

regime, the absorption is controlled by the diffusion of the species to a reaction plane

where their concentrations approach zero – the steepest profile of the absorbed gas is

obtained for this regime. However, for reversible reaction (35) – see Figure 5.5, the

concentration profile at the fiber outlet for IR is smoother than for PFO. This happens

because, at high reaction rates, the reactant depletion becomes significant at a certain

fiber length and the reverse reaction starts playing a role on the process. Close to the

fiber inlet, the steepest concentration profile still corresponds to the IR and, therefore,

the average absorption flux is still higher for this case.

5.3.2. Performance of a membrane contactor for CO2 removal from

anaesthesia breathing circuits

During anaesthesia, the gas flows from the patient at a flow rate, gQ , of approximately

5 -1L min⋅ and with a 2CO concentration of about 5 % (Billiet and Burchill, 2008;

Nguyen, 1996). This 2CO composition must be reduced to a maximum value of 0.5 %

before being recycled back to the patient (Lagorsse et al., 2007).

For this specific application, composite membranes made by a porous support (PEI),

coated internally with a PDMS thin layer were considered. In such membranes, the

main resistance to mass transfer is originated by the PDMS coating. The mass transfer

coefficient of 2CO in this layer, 2,m COk , is approximately 3 -11.58 10 m s−× ⋅ (Rego and

Mendes, 2004). The fibers internal and external diameters were assumed to be,

respectively, 45.6 10−× and 48.4 10−× m.

Aqueous solutions of potassium glycinate were considered as absorbent. The physical

and chemical parameters used to model the system 2CO - potassium glycinate – water

are given in Table 5.1, together with the respective literature sources. Portugal et al.

(2007) derived an expression for the rate of absorption of 2CO in solutions using

157

estimated values of the diffusion coefficients of 2CO and potassium glycinate in

solutions, 2 ,CO SolD and

2 ,RNH SolD . Later, Hamborg et al. (2008) determined these

diffusion coefficients experimentally. The work by Portugal et al. (2007) was re-

analysed using these values and a new expression for the reaction rate was obtained –

Table 5.1.

Given the above considerations, to meet the target of reducing the 2CO concentration

from 5 % to a maximum of 0.5 %, the following design and operation variables can still

be adjusted: the contactor length and shell diameter, the packing density, the liquid

flow-rate, LQ , and the inlet amino acid salt concentration. The influence of these

variables on the system behaviour will be further discussed. The gas phase mass transfer

resistance was neglected and all simulations were performed for a temperature of 298 K

and for fully developed laminar flow in the liquid. 2CO was considered the only

absorbing gas. The conditions studied in the simulations are summarized in Table 5.2.

Given the specificities of the problem under study, namely the fact that the known

restrictions are defined in terms of flow rates rather than velocities – notice that

( )2 1g

shell

Qu

Rπ ε=

− and

2L

shell

Qv

Rπ ε= - equations (29) and (32) were re-written, for

stationary state, as follows:

( )* 2

, *,4

i A ext shelli m

g

u y k R Lf N

x d Q

επ∂=

∂ (37)

2

* *2 2, ,* * *

2 ,2

14i L i Lshell shellA

i RNH feed iL L

C CR L R LDv D k C S

x d Q Q

επ επρρ ρ ρ ∂ ∂∂= − ∂ ∂ ∂

(38)

Equations (30) and (33) can be re-written as well.

158

Table 5.1 - Physical and chemical parameters used to model the absorption of 2CO in potassium glycinate aqueous solutions in a hollow fiber

membrane contactor - liquid phase concentrations are expressed in molarity.

2

2

2

,

,

log CO wRNH

CO Sol

mKC

m

=

( )62.1831 0.1112K T= − ( )-1M (Portugal et al., 2007)

( )2

7, 3.54 10 exp 2044CO wm RT T−= × (Versteeg and Van Swaaij, 1988)

2

2

2 2

-3 -5 2

9, -2 2 -3

-2.412 -9.403 10 7.110 10 -0.2177 -10

-5.447 10 1.296 10

RNH

RNH Sol

RNH RNH

T T CD

C TC−

× + × = × × + ×

( )2 -1m s⋅ (Hamborg et al., 2008)

( )2 2

0.48

, ,CO Sol CO w Sol wD D η η −= ( )2 -1m s⋅ (Hamborg et al., 2008)

( )2

6, 2.35 10 exp 2119CO wD T−= × − ( )2 -1m s⋅ (Versteeg and Van Swaaij, 1988)

( )2 2

21 0.2109 0.05124Sol w RNH RNHC Cη η = + + (Portugal et al., 2007)

( )2

162

86363.28 10 exp exp 0.36RNHk C

T

− = ×

( )-1 -1M s⋅ (Portugal et al., 2007)

2COov

AmA carb

KK

K K= ( )-1M

( )2exp 0.000237956 0.202203 61.6499AmAK T T= − + − ( )M (Perrin, 1965)

( )2

exp 12092.1 36.7816ln 235.482COK T T= − − + ( )M (Benamor and Aroua, 2005)

( )exp 1792 4.786carbK T= − ( )M (Portugal et al., 2009)

158

159

Table 5.2 - Simulation conditions.

( )KT 298.15

( )PaP 105

( )-1, L ming feedQ ⋅ 5

2 ,CO feedy 0.05 (5 %)

( )2

3 -1, 10 m sm COk × ⋅ 1.58

( )410 minnerR × 2.8

ε 0.098 – 0.392

( )210 mshellR × 2 – 4

( )210 mL× 5 – 50

( )-1mL minLQ ⋅ 10 – 500

( )2 , MRNH feedC 0.1 – 3

The influence of packing density, ε , contactor length, L , shell radius, shellR , reactant

feeding concentration, 2 ,RNH feedC , and liquid flowrate, LQ , on the 2CO molar fraction

exiting the contactor is discussed next.

Influence of the packing density, contactor length and shell diameter

The packing density and the contactor volume define the contact area - 22shell

inner

A R LR

ε=

- that is the area available for the mass transfer. Therefore, being gQ and innerR fixed, for

a given LQ and 2 ,RNH feedC , the gas concentration exiting the contactor depends on the

product 2shellR Lε , which is clear from equations (37) and (38) and was also verified by

simulation. The contact area, A, is only limiting the separation process if it is smaller

than the “effective area”, commonly expressed in terms of the effective length at which

the solution becomes saturated and no further separation is achieved. This concept is

illustrated in Figure 5.6, where some representative examples of gaseous concentration

profiles along the contactor are plotted for co- and counter-current operation.

160

Figure 5.6 – Axial profiles along the contactor for co- and counter-current operation and

for different LQ and 2 ,RNH feedC . Simulation conditions: 0.196ε = , Rshell

= 2 × 10−2 m .

It can be seen from this figure that, for the case of a 0.5 M solution flowing at 10

-1mL min⋅ , all the changes in the gas composition occur at a small fraction of the

contactor length (approximately within 12× 10−2 m ) which is the portion effectively

used for the separation. From that point on, increasing the contactor length won’t bring

any advantage for the separation, because the solution is already saturated and unable to

absorb more 2CO . When a 3 M solution is used at the same flow rate, the effective

length increases because this solution has a higher capacity and then saturates later.

Observing this case, one can also notice that the effective length is larger for counter-

current (around 235 10 m−× ) than for co-current operation (around 229 10 m−× ). For both

0.5 and 1 M solutions flowing at 50 -1mL min⋅ , solution capacity is not reached within

the available contact area (this is particularly visible for counter-current operation),

which means that increasing the fiber length would further improve the separation.

Figure 5.6 makes also clear the advantages of operating counter-currently. Counter-

current operation not only enables a more effective use of the contactor area but leads to

161

lower 2CO gas molar fractions exiting the contactor, for the same flow rates and reactant

concentrations.

In Figure 5.7, the gas outlet composition is shown as a function of the contact area for

counter-current operation.

Figure 5.7 - Influence of the contact area on the 2CO molar fraction at the contactor exit

for different 2 ,RNH feedC and LQ and for counter-current operation.

As already observed in Figure 5.6 for the 0.5 M solution, increasing the liquid flow rate

increases the effective area. As a consequence, the separation becomes controlled by the

contact area available – see for example the results for 0.1 M and 500 -1mL min⋅ in

Figure 5.7. Figure 5.7 illustrates more clearly the effect of increasing the amino acid salt

concentration in the contactor effective area.

It can also be observed from Figure 5.7 that, for certain reactant concentrations and

liquid flow rates, it is possible to get outlet 2CO concentrations far below the minimum

required (0.5 %) even when the separation is still limited by the available contact area.

This might be relevant if volume constraints are imposed for the anaesthesia machine.

162

Concerning the contactor design, it must be kept in mind that the value of ε is

constrained by the device geometry. For packing of circles in a plane, the highest

possible packing is 0.907ε = (Weisstein, 2008) which, taking into account the

membrane thickness of 140 µm, corresponds to 0.403ε = , based on the fiber internal

diameter. Due to manufacturing restraints, a reasonable maximum packing density for

the type of contactor used should be about 0.3. Moreover, a final contactor design should

consider an appropriate balance between shell diameter and length. A contactor with a

large diameter and short fibers would lead to the presence of dead volumes or dominant

cross flow operation, while a contactor with very long fibers and short diameter would

imply a high pressure drop and would not fit in a common anaesthetic machine.

Influence of the amino acid salt concentration

As can be seen in Table 5.1, the potassium glycinate concentration influences the 2CO

physical solubility (represented by the partition factor), the 2CO and the reactant

diffusion coefficients and the reaction kinetic constant. The effect of 2 ,RNH feedC on the

2CO molar fraction at the contactor exit is shown in Figure 5.8.

A maximum concentration of 3 M was considered for the simulations. Although in

principle higher concentrations could enhance the separation (as Figure 5.8 indicates),

enabling the use of smaller contactors and lower liquid flow rates, they would also

increase the solution viscosity. Highly viscous solutions would carry hydrodynamic

problems and pumping would become an issue. Precipitation would also become more

likely for concentrated solutions.

163

Figure 5.8 - Influence of the amino acid salt feed concentration liquid flow rate on the

2CO concentration at the contactor exit for different LQ and for co- and counter-current

operations – 20.8796 mA = . Lines are for improving the read.

Influence of the liquid flow rate

The influence of the liquid flow rate on the 2CO molar fraction at the contactor exit is

shown in Figure 5.9.

Increasing the liquid flow rate, LQ , up to a certain value leads to an increase of the 2CO

removal and a consequent decrease on the 2CO gas phase exit molar fraction. However,

for sufficiently high liquid flow rates, other parameters such as the contact area and the

membrane resistance are limiting the 2CO removal. For the asymptotic situation where

there is no depletion of the reactant in the fiber axis along the entire fiber length, the

analogy to conventional mass transfer models is valid. If the conditions for PFO are

fulfilled, the enhancement factor becomes equal to the Hatta number (Danckwerts, 1970)

and therefore the absorption flux is independent of the mass transfer coefficient, Lk - see

164

equations (15) and (16). Since Lk is related to the liquid velocity, increasing LQ in the

PFO regime, for constant centreline reactant concentration, would bring no improvement

for the separation achieved. This effect is visible in Figure 5.9, for counter-current and

for LQ higher than 250 -1mL min⋅ , for concentrations above 0.5 M.

Figure 5.9 – Influence of the liquid flow rate on the 2CO concentration at the contactor

exit for different 2 ,RNH feedC and for co- and counter-current operations – 20.8796 mA = .

Lines are for improving the read.

Figure 5.9 shows that, for a number of operating conditions, the 2CO outlet

concentration is well below the limit for anaesthesia purposes. One must keep in mind

that, although this work is focused on the absorption step, the global process also

comprises the continuous regeneration of the solution. Usually the 2CO desorption is

carried out upon heating the rich solutions, which makes the regeneration step critical in

what concerns energy consumptions. Therefore, to minimize the energy requirements,

the liquid flow rate should be the lowest possible. As a consequence, the reactant feed

concentration should be the highest possible, which, given the above considerations

corresponds to 3 M.

165

According to Figures 5.7 to 5.9, a contactor having 0.5 m2 of contact area, with a 3 M

aqueous solution of potassium glycinate, flowing at 10 -1mL min⋅ counter-currently with

respect to the gas would be suitable to reduce the 2CO molar fraction in anaesthesia

closed loop from 5 % to less than 0.5 %. Considering a packing density of 0.3, a

contactor for this application should have a volume of approximately 240 mL. Taking

into consideration the dimensions of the usual alkali hydroxides canisters, a contactor

with 5 cm shell diameter and 12.5 cm length would be easily retrofitted into a common

anaesthesia machine.

5.4. Conclusions

A numerical model was developed to simulate the mass transfer accompanied by

chemical reaction, occurring during the absorption of a gas into a liquid flowing through

a hollow fiber membrane contactor. The model considers the liquid flowing in the fiber

lumen and the gas in the shell side. Both co- and counter-current operations were

analysed. Good agreement was found between the proposed numerical model results and

results obtained with conventional mass transfer models, for the limit conditions where

these are valid.

The performance of a hollow fiber membrane contactor with potassium glycinate

absorbent solutions was studied for the purpose of carbon dioxide removal from

anaesthesia closed-loops. The effect of the contact area on the 2CO molar fraction

exiting the contactor was analysed; the contact area is defined by the design parameters

packing density, contactor length and shell diameter. The separation achieved increases

by increasing the contact area until the liquid solution becomes saturated, which happens

at the so-called effective area or effective length. The operation parameters liquid flow

rate and reactant feed concentration determine the area effectively used for the mass

transfer.

The effect of the reactant feed concentration on the separation was examined and, for the

concentration range considered (0 to 3 M), the amount of 2CO absorbed always

increased with increasing concentration.

166

Generally, increasing the liquid flow rate increases the separation. However, when the

contactor is working in the limit situation of PFO and no depletion of the reactant at the

fiber axis, the liquid velocity has no effect on the absorption rate and consequently on the

2CO molar fraction exiting the contactor. Concerning the application under study, to

minimize energy requirements for the thermal desorption, one are interested on keeping

the liquid flow rate the lowest possible.

Finally, based on practical considerations and restrictions, a set of operating conditions

and design parameters were assumed for the contactor and its viability confirmed by

simulation. These are a packing density of 0.3 and a volume of 240 mL (approximately

0.5 m2 of contact area), with a 3 M solution of potassium glycinate flowing counter

currently with respect to the gas, at a flow rate of 10 -1mL min⋅ .

5.5. Nomenclature

A Contact area, 2m

C Concentration, M or -3mol m⋅

d Fiber internal diameter, m


Da Damköhler number, dimensionless



f Flag standing for co- or counter-current operation

Gz Graetz number, dimensionless


IR Instantaneous reaction regime

J Absorption flux, -2 -1mol m s⋅ ⋅

ovK Overall equilibrium constant

1k− Reverse reaction kinetic constant, s-1

2k Second order reaction kinetic constant, M-1 ⋅s-1

extk External mass transfer coefficient, -1m s⋅

167

gk Gas phase mass transfer coefficient, -1m s⋅

mk Membrane mass transfer coefficient, -1m s⋅


ovk Overall mass transfer coefficient, -1m s⋅

L Fiber length, m

m Partition coefficient

n Number of fibers

,i mN Flux across the membrane, -2 -1mol m s⋅ ⋅

nj Number of discretization points in the axial direction

nk Number of discretization points in the radial direction

PFO Pseudo first order reaction regime

Q Flow rate, -1L min⋅ or -1mL min⋅

r Radial coordinate, m


innerR Membrane internal radius, m

/L GRτ Ratio between liquid and gas residence times, dimensionless

/L MRτ Ratio liquid residence and membrane characteristic times, dimensionless

shellR Contactor shell inner radius, m

S Rate of reaction, -3 -1mol m s⋅ ⋅

Sh Sherwood number, L

A

k dSh

D= , dimensionless

T Temperature, K

u Gas velocity, -1m s⋅

v Liquid velocity, -1m s⋅

V Volume, m3

x Axial coordinate, dimensionless

z Axial coordinate, m

Greek symbols

ϕ Flux, -2 -1mol m s⋅ ⋅

ε Packing density, dimensionless

168

ν Stoichiometric coefficient


ρ Radial coordinate, dimensionless

θ Time, dimensionless

τ Residence time, s

Subscripts

0 Initial

∞ Infinite (instantaneous reaction regime)

A Absorbed gas

B Non-volatile reactant

,C D Reaction products

F Cell face

g Gas phase

i Gaseous component

int Interface

L Liquid phase

m Membrane

ref Reference

Sol Solution

T Total

w Water

Superscripts

* Dimensionless

z Axial coordinate, m

r Radial coordinate, m

5.6. References

Al-Marzouqi, M., El-Naas, M., Marzouk, S. and Abdullatiff, N. (2008). "Modeling of

chemical absorption of CO2 in membrane contactors." Separation and Purification

Technology, 62(3), 499-506.

169

Bao, L. H. and Trachtenberg, M. C. (2005). "Modeling CO2-facilitated transport across a

diethanolamine liquid membrane." Chemical Engineering Science, 60(24), 6868-6875.

Baum, J. A. and Woehlck, H. J. (2003). "Interaction of inhalational anaesthetics with

CO2 absorbents " Best Practice and Research Clinical Anaesthesiology, 17(1), 63-76.

Benamor, A. and Aroua, M. K. (2005). "Modeling of CO2 solubility and carbamate

concentration in DEA, MDEA and their mixtures using the Deshmukh-Mather model."

Fluid Phase Equilibria, 231(2), 150-162.

Billiet, P. and Burchill, S. (2008). "The Open Door Web Site - Breathing in the Air: The

Lungs - http://www.saburchill.com." 2008.

Bird, R. B., Sewart, W. E. and Lightfoot, E. N. (2002). "Transport phenomena". New

York, John Wiley and Sons, inc.

Blauwhoff, P. M. M., Versteeg, G. F. and Vanswaaij, W. P. M. (1984). "A Study on the

Reaction between Co2 and Alkanolamines in Aqueous-Solutions." Chemical


Boucif, N., Favre, E. and Roizard, D. (2008). "CO2 capture in HFMM contactor with

typical amine solutions: A numerical analysis." Chemical Engineering Science, 63(22),

5375-5385.



Chen, G., Ren, Z. Q., Zhang, W. D. and Gao, J. (2007). "Modeling study of the influence

of porosity on membrane absorption process." Separation Science and Technology,

42(15), 3289-3306.

Coker, D. T., Freeman, B. D. and Fleming, G. K. (1998). "Modeling multicomponent gas

separation using hollow-fiber membrane contactors." American Institute of Chemical


Courant, R., Isaacson, E. and Rees, M. (1952). "On the Solution of Nonlinear Hyperbolic

Differential Equations by Finite Differences." Communications on Pure and Applied

Mathematics, 5(3), 243-255.

170

Cruz, P., Santos, J., Magalhaes, F. and Mendes, A. (2005). "Simulation of separation

processes using finite volume method." Computers & Chemical Engineering, 30(1), 83-

98.

Danckwerts, P. (1970). "Gas-Liquid Reactions", McGraw-Hill Book Company.

DeCoursey, W. J. (1974). "Absorption with Chemical-Reaction - Development of a New

Relation for Danckwerts Model." Chemical Engineering Science, 29(9), 1867-1872.

Derks, P. W. J., Kleingeld, T., van Aken, C., Hogendoom, J. A. and Versteeg, G. F.

(2006). "Kinetics of absorption of carbon dioxide in aqueous piperazine solutions."


Dindore, V., Brilman, D. and Versteeg, G. (2005a). "Hollow fiber membrane contactor

as a gas-liquid model contactor." Chemical Engineering Science, 60(2), 467-479.

Dindore, V., Brilman, D. and Versteeg, G. (2005b). "Modelling of cross-flow membrane

contactors: Mass transfer with chemical reactions." Journal of Membrane Science,

255(1-2), 275-289.

Elk, E. P. V., Knaap, M. C. and Versteeg, G. F. (2007). "Application of the penetration

theory for gas-liquid mass transfer without liquid bulk: Differences with systems with a

bulk." Chemical Engineering Research & Design, 85(4), 516-524

Fan, S. Z., Lin, Y. W., Chang, W. S. and Tang, C. S. (2008). "An evaluation of the

contributions by fresh gas flow rate, carbon dioxide concentration and desflurane partial

pressure to carbon monoxide concentration during low fresh gas flows to a circle

anaesthetic breathing system." European Journal of Anaesthesiology, 25(8), 620-626.




Gabelman, A. and Hwang, S. (1999). "Hollow fiber membrane contactors." Journal of


171




Hamborg, E. S., Niederer, J. P. M. and Versteeg, G. F. (2007). "Dissociation constants

and thermodynamic properties of amino acids used in CO2 absorption from (293 to 353)

K." Journal of Chemical and Engineering Data, 52(6), 2491-2502.

Hamborg, E. S., van Swaaij, W. P. M. and Versteeg, G. F. (2008). "Diffusivities in

aqueous solutions of the potassium salt of amino acids." Journal of Chemical and


Hoff, K. A., Juliussen, O., Falk-Pedersen, O. and Svendsen, H. F. (2004). "Modeling and

experimental study of carbon dioxide absorption in aqueous alkanolamine solutions

using a membrane contactor." Industrial & Engineering Chemistry Research, 43(16),

4908-4921.

Hogendoorn, J. A., Vas Bhat, R. D., Kuipers, J. A. M., van Swaaij, W. P. M. and

Versteeg, G. F. (1997). "Approximation for the enhancement factor applicable to

reversible reactions of finite rate in chemically loaded solutions." Chemical Engineering

Science, 52(24), 4547-4559.



36(5), 1779-1790.

Idem, R. and Tontiwachwuthikul, P. (2006). "Preface for the special issue on the capture

of carbon dioxide from industrial sources: Technological developments and future

opportunities." Industrial & Engineering Chemistry Research, 45(8), 2413-2413.

Keshavarz, P., Ayatollahi, S. and Fathikalajahi, J. (2008). "Mathematical modeling of

gas–liquid membrane contactors using random distribution of fibers." Journal of

Membrane Science, 325, 98–108.

Knaap, M. C., Lenferink, J. E. O., Versteeg, G. F. and van Elk, E. P. (2000).

"Differences in local absorption rates of CO2 as observed in numerically comparing tray

172

columns and packed columns." 79th Annual Convention of the Gas Processors

Association, Proceedings, 82-94

596.

Knolle, E. and Gilly, H. (2000). "Absorption of carbon dioxide by dry soda lime

decreases carbon monoxide formation from isoflurane degradation." Anesthesia and

Analgesia, 91(2), 446-451.

Kreulen, H., Smolders, C. A., Versteeg, G. F. and Vanswaaij, W. P. M. (1993a).

"Microporous Hollow-Fiber Membrane Modules as Gas-Liquid Contactors .1. Physical

Mass-Transfer Processes - a Specific Application - Mass-Transfer in Highly Viscous-

Liquids." Journal of Membrane Science, 78(3), 197-216.

Kreulen, H., Smolders, C. A., Versteeg, G. F. and Vanswaaij, W. P. M. (1993b).

"Microporous Hollow-Fiber Membrane Modules as Gas-Liquid Contactors .2. Mass-

Transfer with Chemical-Reaction." Journal of Membrane Science, 78(3), 217-238.

Kumar, P., Hogendoorn, J., Versteeg, G. and Feron, P. (2003a). "Kinetics of the reaction

of CO2 with aqueous potassium salt of taurine and glycine." American Institute of

Chemical Engineers Journal, 49(1), 203-213.

Kumar, P., Hogendorn, J., Feron, P. and Versteeg, G. (2003b). "Approximate solution to

predict the enhancement factor for the reactive absorption of a gas in a liquid flowing

through a microporous membrane hollow fiber." Journal of Membrane Science, 213(1-

2), 231-245.

Kumar, P. S. (2002). "Development and design of membrane gas absorption processes".

Enschede, University of Twente.

Kumar, P. S., Hogendoorn, J. A., Feron, P. H. M. and Versteeg, G. F. (2002). "New

absorption liquids for the removal of CO2 from dilute gas streams using membrane

contactors." Chemical Engineering Science, 57(9), 1639-1651.

Kumar, P. S., Hogendoorn, J. A., Feron, P. H. M. and Versteeg, G. F. (2003c).


Crystallization in carbon dioxide loaded aqueous salt solutions of amino acids."

Industrial & Engineering Chemistry Research, 42(12), 2832-2840.

173

Kumar, P. S., Hogendorn, J. A., Feron, P. H. M. and Versteeg, G. F. (2003d).

"Approximate solution to predict the enhancement factor for the reactive absorption of a

gas in a liquid flowing through a microporous membrane hollow fiber." Journal of


Lagorsse, S., Magalhaes, F. D. and Mendes, A. (2007). "Xenon recycling in an

anaesthetic closed-system using carbon molecular sieve membranes." Journal of


Li, J. L. and Chen, B. H. (2005). "Review Of CO2 absorption using chemical solvents in

hollow fiber membrane contactors." Separation and Purification Technology, 41(2), 109-

122.

Ma'mun, S., Svendsen, H. F., Hoff, K. A. and Juliussen, O. (2007). "Selection of new

absorbents for carbon dioxide capture." Energy Conversion and Management, 48(1),

251-258.

Majchrowicz, M., Niederer, J. P. M., Velders, A. H. and Versteeg, G. F. (2006).

"Precipitation in amino acid salt CO2 absorption systems". GHGT-8 NTNU VIDERE,

Pav. A, Dragvoll, NO-7491 Trondheim, Norway.




163.

Nguyen, J. (1996). "End-tidal carbon dioxide monitoring during CPR: A predictor of

outcome - http://enw.org/ETCO2inCPR.htm."

Paul, S., Ghoshal, A. K. and Mandal, B. (2007). "Removal of CO2 by single and blended

aqueous alkanolamine solvents in hollow-fiber membrane contactor: Modeling and

simulation." Industrial & Engineering Chemistry Research, 46(8), 2576-2588.

Perrin, D. (1965). "Dissociation Constants of Organic Bases in Aqueous Solutions".

London, Butterworth.

Petzold, L. R. and Hindmarsh, A. C. (1997). LSODA, Computer and mathematics

research division, Lawrence Livermore National Laboratory.

174

Portugal, A. F., Derks, P. W. J., Versteeg, G. F., Magalhães, F. D. and Mendes, A.

(2007). "Characterization of potassium glycinate for carbon dioxide absorption purposes

" Chemical Engineering Science, 62(23), 6534-6547

Portugal, A. F., Magalhaes, F. D. and Mendes, A. (2008). "Carbon dioxide absorption

kinetics in potassium threonate." Chemical Engineering Science, 63(13), 3493-3503.

Portugal, A.F., Sousa, J.M., Magalhães, F. D. and Mendes A. (2009). "Solubility of

carbon dioxide in aqueous solutions of amino acid salts", Chemical Engineering Science,

10.1016/j.ces.2009.01.036

Rego, R. and Mendes, A. (2004). "Carbon dioxide/methane gas sensor based on the

permselectivity of polymeric membranes for biogas monitoring." Sensors and Actuators

B-Chemical, 103(1-2), 2-6.

Santos, J. C., Cruz, P., Regala, T., Magalhaes, F. D. and Mendes, A. (2007). "High-

purity oxygen production by pressure swing adsorption." Industrial & Engineering

Chemistry Research, 46(2), 591-599.

Sartori, G. and Savage, D. W. (1983). "Sterically Hindered Amines for Co2 Removal


Sea, B., Park, Y. I. and Lee, K. H. (2002). "Comparison of porous hollow fibers as a

membrane contactor for carbon dioxide absorption." Journal of Industrial and

Engineering Chemistry, 8(3), 290-296.

Secor, R. M. and Beutler, J. A. (1967). "Penetration Theory for Diffusion Accompanied

by a Reversible Chemical Reaction with Generalized Kinetics." American Institute of

Chemical Engineers Journal, 13(2), 365-373.

Song, H. J., Lee, S., Maken, S., Park, J. J. and Park, J. W. (2006). "Solubilities of carbon

dioxide in aqueous solutions of sodium glycinate." Fluid Phase Equilibria, 246(1-2), 1-5.

Sousa, J. M. and Mendes, A. (2003). "Simulation study of a dense polymeric catalytic

membrane reactor with plug-flow pattern." Chemical Engineering Journal, 95(1-3), 67-

81.

175

Supap, T., Idem, R., Tontiwachwuthikul, P. and Saiwan, C. (2006). "Analysis of

monoethanolamine and its oxidative degradation products during CO2 absorption from

flue gases: A comparative study of GC-MS, HPLC-RID, and CE-DAD analytical

techniques and possible optimum combinations." Industrial & Engineering Chemistry

Research, 45(8), 2437-2451.


Complex Reversible Chemical-Reactions in Gas-Liquid Systems - an Overview."



Gases (Co2, N2O) in Aqueous Alkanolamine Solutions." Journal of Chemical and


Wang, R., Li, D. F. and Liang, D. T. (2004). "Modeling of CO2 capture by three typical

amine solutions in hollow fiber membrane contactors." Chemical Engineering and

Processing, 43(7), 849-856.

Weisstein, E. W. (2008). ""Circle Packing." From MathWorld-A Wolfram Web

Resource - http://mathworld.wolfram.com/CirclePacking.htmlWeisstein." 2008.

Whalen, F. X., Bacon, D. R. and Smith, H. M. (2005). "Inhaled anesthetics: an historical

overview." Best Practice Research Clinical Anaesthesiology, 19(3), 323-330.

Zhang, H. Y., Wang, R., Liang, D. T. and Tay, J. H. (2006). "Modeling and experimental

study of CO2 absorption in a hollow fiber membrane contactor." Journal of Membrane

Science, 279(1-2), 301-310.

Zhang, Q. and Cussler, E. L. (1985). "Microporous Hollow Fibers for Gas-Absorption .1.

Mass-Transfer in the Liquid." Journal of Membrane Science, 23(3), 321-332.

5.A. Spatial discretization method

Equations in this appendix are written in their dimensional form to simplify

understanding.

176

The hollow fiber membrane contactor was discretized as shown in Figure 5.A1 - both

shell and fiber are divided in nj equally spaced intervals in the axial direction; the fiber

lumen is also divided in the radial direction following a geometric progression:

1

1

1 0.85

1 0.85

k

k innernkr R

−

−

−=−

(B.1)

where nk is the number of finite volumes in the radial direction. It was found that 16

axial discretization points and 64 radial discretization points were sufficient to describe

the problem.

Figure 5.A1 – Schematic representation of the spatial discretization and cell mass

balance.

For each cell and for each component i present in the system, the following mass

balances can be written:

177

Liquid phase mass balance

( ) ( )

( ) ( )

1 1

2 21 1 1

what enters in radial directionwhat enters in axial direction

2 21 1

what comes what comes out in axial direction

2

2

j k

j k

z rFz k k Fr k j j

z rFz k k Fr k j j

r r r z z

r r r z z

ϕ π ϕ π

ϕ π ϕ π

− ++ + −

+ −

− + − =

= − + −

( )( ) ( )( )out in radial direction

,2 2 2 21 1 1 1

what disapears upon reactionwhat accumulates in the element

i liqk k j j k k j j i

dCr r z z r r z z S

dtπ π+ − + −

+

+ − − + − −

(B.2)

According to the model assumptions, zFϕ and rFϕ , respectively, the axial and radial

fluxes crossing the cell faces, are given by , ,zF F i L Fv Cϕ = and ,i Lr

F i

F

dCD

drϕ = . Then,

substituting in (B.2), the following equation results:

( ) ( )

1 1 1

, ,1

, , , ,,

2 21 1

2j j j j k k

i L i Lk k

Fz i L Fz Fz i L Fz Fr Fri Li i

j j k k

C Cr r

v C v C r rCD S

t z z r r− − +

+

− +

∂ ∂−

− ∂ ∂∂= − + −

∂ − − (B.3)

where ,i LC is the concentration of i in the cell, 1jFzv

− and

jFzv are the liquid velocities

respectively in the upstream and downstream faces of the cell - in case of laminar flow

( ) ( )2 2

12 12j

k kFz

r R r Rv v +

+= −

, 1, , ji L FzC

− and , , ji L FzC are the concentrations of i

respectively in the upstream and downstream faces of the cell and 1

,

k

i L

Fr

C

r+

∂∂

and ,

k

i L

Fr

C

r

∂∂

are the concentration gradients, respectively in the outer and inner faces of the liquid

cylindrical shell. Initial and axial boundary conditions of equation (B.3) are

straightforwardly implemented: , , ,00, i L i Lt C C= = and 1, , , ,0, i L Fz i L feedz C C= = .

Concerning the radial boundary condition for the absorbing compounds, an infinitesimal

volume, infV , with concentration ,i mC , was created in the fiber wall and a mass balance

was performed in this volume:

( )1, ,

2

nk

j ji m i Lim i

inf Fr

R z zC CN D

t V r

π − −∂ ∂ = − ∂ ∂

(B.4)

where the flux through the membrane, ,i mN , is given by: ( ), , , ,i m i ext i g i mN k C C= − . The

radial boundary condition of equation (B.3) then becomes: , , ,,nkinner i L Fr i i mr R C m C= =

178

and the initial condition of equation (B.4) was set as: , , ,00, i m i L it C C m= = . This

strategy considerably attenuates the numerical instability associated to the condition of

fluxes equality in the interface.

Gas phase mass balances were computed equivalently, but only axial flux was

considered ( ), ,zF F i g Fu Cϕ = .

Part V

181

General Conclusions and Future Work

The present work aimed at studying the use of hollow fiber membrane contactors for

2CO removal from anaesthetic gas streams by selective absorption.

The 2CO removal from closed anaesthetic loops is currently achieved using mixtures of

alkali hydroxides which, under desiccated conditions, react with the anaesthetic

volatiles originating highly toxic compounds. In addition, this technique is associated to

explosions due to the hydrogen formation and excessive heating during the reaction of

these absorbents with 2CO . Furthermore, the exhaust containers of the absorbent

mixtures are hospital solid waste. These reasons drive the need for replacing this

absorption system with a safer and more environmentally friendly technology. The use

of hollow fiber membrane contactors with renewable liquid absorbents is a possible

strategy to overcome most of the pointed out drawbacks. Using dense and highly

permeable membranes in such devices, the absorption system can be kept isolated from

the anesthetic loop and aseptic operation is possible. The absorbent solution can be

subsequently regenerated after contacting with 2CO .

Most of the present work was focussed on the study of absorbent solutions suitable for

the intended application. Such solution must be biocompatible, chemically and

thermally stable and have a low vapour pressure. Additionally, it must present high

absorption and desorption kinetics and high absorption capacity and should be easy to

regenerate. Aqueous solutions of alkali salts of amino acid are expected to fulfil these

requirements and, therefore, two amino acid salts were characterized for 2CO

absorption: potassium glycinate (because glycine is the simplest amino acid, it has a

relatively low cost and its molecular structure indicates high absorption kinetics) and

potassium threonate (because its molecular structure envisioned better regeneration

properties).

Following the absorbents characterization, the performance of the contactor for 2CO

removal from closed-loop anaesthesia was analysed by simulation, using the physico-

chemical properties obtained for potassium glycinate. The use of hollow fiber absorbent

182

membrane contactors for the 2CO removal from closed anaesthetic breathing circuits

was proposed for the first time by our research group. The analysis performed in the

present dissertation indicates that this technology is suitable. The absorbent regeneration

process, however, still has to be studied in some detail.

Some relevant results obtained along the different phases of the work are described

below.

Physical Properties Measurements

Densities and viscosities of potassium glycinate and potassium threonate aqueous

solutions were measured for amino acid salt concentrations ranging from 0.1 to 3.0 M

and temperatures from 293 to 313 K. For the concentration range analysed, the increase

of density and viscosity are not likely to bring additional hydrodynamic concerns or

pumping difficulties. However, for potassium threonate concentrations above 3.0 M and

low temperatures, the solution viscosity might become too high for use in hollow fibre

membrane modules.

Diffusion coefficients of 2N O and the amino acids salts in amino acid salts solutions

were estimated using the modified Stokes Einstein relation. 2CO diffusion coefficients

were computed using the 2N O analogy.

To estimate the physical solubility of 2CO in amino acid salts solutions, 2N O solubility

was measured and the results interpreted using the Shumpe model. These measurements

were performed for potassium glycinate and potassium threonate concentrations from

0.1 to 3.0 M and temperatures from 293 to 313 K.

Adsorption Kinetics Studies

The kinetics of the reactions of 2CO with potassium glycinate and potassium threonate

were determined using a stirred cell working semi-continuously with respect to the gas

phase and batchwise with respect to the liquid phase. It was concluded that both amino

183

acid salts presented absorption rates towards 2CO similar to alkanolamines. The results

also indicate that the reaction rate significantly depends on the ionic strength of the

solution. As expected, because of the molecular configuration, it was verified that

potassium glycinate shows a faster absorption of 2CO than potassium threonate.

The enhancement factor, and subsequently the overall kinetic constant, was computed

using the DeCorsey equation. The apparent kinetic constant of potassium glycinate is in

line with the Brønsted plot drawn for other amines, whereas the apparent kinetic

constant for potassium threonate falls below the plot. This is likely to be due to the

sterical hindrance of the amine group from threonate.

For potassium glycinate, the rate of absorption as a function of temperature and amino

acid salt concentration, for the conditions studied, was found to be given by the

following expression: ( )2 2


T

− − = ×

-3 -1mol m s⋅ ⋅ . Based on experimental diffusivity data recently published, this expression

was later updated: ( )2 2


T

− − = ×

-3 -1mol m s⋅ ⋅ .

The absorption rate as a function of the temperature and concentration found for

potassium threonate was: ( )2 2


T

− − = ×

-3 -1mol m s⋅ ⋅ .

Adsorption Equilibrium Studies

2CO solubility in potassium glycinate aqueous solutions with concentrations from 0.1 to

3.0 M and temperatures from 293 to 351 K was determined in a stirred cell. 2CO

solubility in a 1.0 M solution of potassium threonate was also measured at 313 K.

Potassium glycinate showed absorption capacities towards 2CO (expressed in terms of

2

1CO AmAmol mol−⋅ ) similar to monoethanolamine. On the other hand, potassium threonate

showed a considerably lower 2CO absorption capacity. This can be assigned to the

184

sterical hindrance of the threonate amine group (higher carbK ) along with the lower p AK

(then higher AmAK ).

Potassium glycinate did not show significant change on the 2CO solubility for

temperatures between 293 and 323 K. This result is somehow surprising and is a sign of

potential problems in the regeneration of the absorbent solutions.

The Deshmukh-Mather and the Kent-Eisenberg models were used to interpret the

equilibrium results. Although the predictions of both models significantly deviate from

the experimental results, they prove to qualitatively describe the system. These models

are particularly useful to provide the composition of the solutions (speciation) as a

function of loading and temperature. Speciation enables the prediction of 2CO

absorption rates in partially loaded solutions likely to take place in

absorption/desorption cycles.

No precipitation was observed for any of the solutions of potassium glycinate and

potassium threonate studied, which indicates that these are suitable for hollow fiber

membrane contactors, even if porous membranes are to be used.

Hollow Fiber Membrane Contactor Simulation

A coupled differential model for both gas and liquid phases was proposed to simulate

the mass transfer accompanied by chemical reaction, occurring during the absorption of

2CO into amino acid salt solutions flowing through a hollow fiber membrane contactor.

The model considered the liquid flowing in the fiber lumen and the gas in the shell side.

Both co- and counter-current operations were considered.

The developed model enables to assess the radial and axial concentration profiles in the

liquid and the axial velocity and concentration profiles in the gas.

Model results were compared to results obtained with conventional mass transfer

models valid for limit conditions and good agreement was found.

185

The performance of a hollow fiber membrane contactor for 2CO removal from

anaesthesia closed-loops was analysed. For this analysis, composite PDMS membranes

were assumed and aqueous solutions of potassium glycinate were considered as

absorbents. The physical, kinetic and equilibrium data experimentally obtained for

potassium glycinate were used in the simulations.

The influence of the design parameters packing density, contactor length and shell

diameter and of the operation parameters liquid flow rate and reactant feed

concentration, on the 2CO molar fraction exiting the contactor were studied. Based on

this study, a contactor with 5 cm shell diameter and 12.5 cm length, with a 3 M solution

of potassium glycinate flowing at 10 -1mL min⋅ , counter-currently with respect to the

gas was found to be suitable for the 2CO removal from closed anaesthetic breathing

circuits. A contactor with these dimensions would be easily retrofitted into a common

anaesthesia machine (note that common soda lime canisters have 1.5 L).

Suggestions for Future Work

Literature information about amino acid salts as 2CO absorbents is still scarce, although

it keeps increasing due to their potential application to flue gas treatment. A more

comprehensive data set is needed, namely concerning physical properties (such as

diffusion coefficients and physical solubility) and equilibrium and kinetics data. This

will allow for a more consistent implementation of the data analysis proposed in the

present work. Furthermore, it may serve as a basis for more complex and accurate

models.

In particular, there is a lack of information in the open literature concerning the

regeneration of the absorbent solutions. After absorbing 2CO , the solution should be

regenerated (possibly in another device) and used again for absorption. Therefore, it is

the cyclic performance of the solution which will define the viability of the global

process. Additionally, the regeneration is usually performed at high temperatures,

186

making this step critical in what concerns energy consumption. For these reasons,

research on the regeneration subject is suggested for future work.

Pursuing this suggestion, measurements of all properties at higher temperatures are

strongly recommended, since the studies of the 2CO absorption in amino acid salts

performed so far were limited to a relatively narrow temperature interval when

compared to the conditions expected to be found in cyclic absorption/regeneration

processes. Moreover, a method to determine the 2CO desorption kinetics should be

developed.

Concerning the specific application studied in the present work, 2CO removal from

anaesthetic gaseous circuits, it is suggested to perform multi component experiments

with gaseous mixtures similar to the ones present in real anaesthesia. In particular, these

mixtures should include the halogenated anaesthetics, in order to check the effect of

these compounds in the absorbent solutions as well as in the membrane materials.

Finally, the complete cyclic process should be simulated, experimentally validated and

further optimized for the required separations.

Appendix A

Details on the Experimental Setups Used

188

Details on the Experimental Setups Used

Some considerations on the setups used in the experimental work, which are not detailed

in the papers, are presented next.

All the experiments concerning potassium glycinate (apart the viscosity measurements)

were performed in the OOIP Group of the Department of Development and Design of

Industrial Processes in the Twente University, Enschede, The Netherlands. The setup

used for the physical solubility and kinetics measurements is presented in Figures A1

and A2. Descriptions of the experimental procedures are presented in Chapter 2.

Figure A1 – Setup used for the physical absorption and kinetics measurements of 2CO in

potassium glycinate (Chapter 2) - the gas vessel and pressure controller are located

behind the panel.

Gas in

Degassing and physical

solubility tank

Stirred Reactor

189

Figure A2 – Detail of the setup - stirred reactor (liquid volume: 600 cm3, reactor

diameter: 9.09 cm).

For the characterization of potassium threonate (Chapter 3), an apparatus with similar

functioning but with smaller dimensions (about 12 times smaller than the one used for

Chapter 2) was built in LEPAE - Laboratory for Process, Environmental and Energy, in

the Department of Chemical Engineering from the University of Porto. The setup,

designed and assembled by the author of this thesis, is shown in Figure A3.

The main advantage of using a reactor with smaller dimensions is that less reactant is

necessary for the experimental measurements. This makes the characterization more

economic and pursues the objective of characterizing non-commercially available

absorbents (although around 1 kg of absorbent is still needed for a reasonable

characterization). On the other hand, eventual experimental error sources (including

leaking, dead volumes, presence of solution drops in the reactor walls, among others)

have a much more noticeable effect on experimental results. To check the accuracy of

the setup assembled, results obtained for a 1 M solution of potassium glycinate at 298 K

were compared with these obtained with the setup used in Chapter 2. Figure A4 shows

the absorption flux as a function of the 2CO partial pressure obtained using the setups

shown in Figures A1 and A3.

Absorbent solution

Gas

Gas In

190

Figure A3 - Setup used for the determination of the physical absorption and reaction

kinetics of 2CO in potassium threonate (Chapter 3) and for the equilibrium

measurements of 2CO in potassium glycinate (Chapter 4) - liquid volume: 50 cm3,

reactor diameter: 3.87 cm.

Stirred reactor

Degassing tank

Gas vessel

Pressure controller

191

0 10 20 30 40

0.0000

0.0005

0.0010

0.0015

0.0020

0.0025

0.0030

0.0035

LEPAE - Porto UniversityOOIP - Twente University

( )2

210 PaCOP −×

( )-1 -1mol m sJ ⋅ ⋅

Figure A4 – Comparison of the experimental results obtained using the setup at Porto

University and using the setup from Twente University.

The setup shown in Figure A3 was also used to perform the experiments presented in

Part III (Chapter 4).

carbon dioxide removal from anaesthetic gas circuits using ... integr… · carbon dioxide removal...

Documents